IQ Trend Beams [TradingIQ]🔹 OVERVIEW
IQ Trend Beams is a trend assistant that draws your trendlines the way a disciplined chartist would - and then holds them accountable. It maintains two channels, support and resistance , each always showing one working line. A line is born forming : it moves and re-shapes freely, polished every bar by a perceptual score toward the line a skilled trader would actually draw. When its geometry settles and it has earned enough tangency credit, it locks - and from that moment the ink is frozen forever; it never moves again. Locked ink extends until break evidence fires, then it is broken : restyled but never relocated, holding the screen as history until its successor locks.
Riding each live beam is its own forecast ; a calibration band, a reach profile, and ghost levels, all built from the volume that has actually traded around that line.
This is an honest visualization and modeling tool , not a signal service. It draws structure clearly and states its own confidence out loud; it is not a validated edge or a promise of profit. Read the limitations section - it is not window dressing.
🔹 THE TWO CHANNELS - AN AUDITED PROMISE
Most trendline tools quietly redraw the past so the line always looks right in hindsight. Trend Beams refuses to. A line lives through three visible states:
• Forming (dotted) - the assistant sketching. It is free to move and re-fit while it hunts for the right geometry. This is the only state in which a support/resistance line moves, and it is dotted precisely so you can tell a guess from a commitment.
• Locked (solid) - the geometry has stilled and earned its tangency credit, so the line is frozen . It will never move again. A locked beam is a promise the tool has to keep in public.
• Broken (restyled) - break evidence fired. The ink is re-styled to show it failed, but it is never relocated ; it holds its original slope as an honest record and, if you keep history on, dims into the background once its successor locks.
Because a locked line cannot move, what you saw at lock time is what you keep. This is the core design commitment of the tool.
Two rails, either direction by design. Support is the lower rail, fit to the swing lows on the underside of price; resistance is the upper rail, fit to the swing highs above it. Neither is locked to a single slope: in a falling market the support rail angles down with the lows (the floor of the down-channel), and in a rally the resistance rail angles up with the highs (the ceiling of the up-channel). That is deliberate. A tool that forces support to only ever point up would go blind to the lower boundary of a downtrend - and miss exactly the moves that matter. Trend Beams instead always draws both boundaries of the channel price is actually in , so a strong move is framed on both sides rather than half-missed. If you prefer to read it the classical way, follow the rail that agrees with the trend and treat the other as the opposite wall of the same channel.
🔸 HOW A LINE EARNS ITS LOCK
While forming, each line is scored every bar by a perceptual fit , a running measure of how well its geometry matches what a careful trader would draw against the recent swing structure, blended with a one-pole toward its fitted slope so it settles rather than twitches. A lock is granted only when the geometry has gone still for long enough, the line has accumulated real tangency credit (genuine touches, not a single graze), and it spans a minimum bar count - and it is refused outright if it would invert the channel. The Mode dial sets how much evidence this takes.
🔹 THE AUDIT BADGE
Locked ink can carry a small measurement badge that reports, in plain terms, how the line is actually holding up:
• Wick-through - recent piercing of the line, exponentially weighted, measured against the tool's 10% design target . A well-behaved line lets price kiss it, not knife through it.
• Survival probability - the current modeled odds that the line is still valid.
• Maturity - how far through its estimated total run the move is, so a young trend reads differently from an exhausted one.
The badge is the tool grading its own work on the chart, not a trade instruction.
🔸 THE FORECAST - EACH BEAM READS ITS OWN VOLUME
Every live beam carries its own forecast, built entirely from the volume that has traded around that line. Trend Beams bins the intrabar volume by its distance from the beam, smooths it into a continuous density (a kernel-density estimate), and renders three things that ride the line:
• Calibration band - translucent ribbons hugging the beam, one per density bin, showing where the trend has held its volume. Strength is encoded as colour vibrancy at a constant perceptual lightness (the Oklab principle - a dense core reads vivid, the thin tails fade), so nothing is made brighter or darker than its weight warrants.
• Reach profile - a smooth filled contour fanning into the future margin, where each level's forward extent is its density times the trend's estimated remaining length . It answers, at a glance: if this trend keeps going, how far - and around which prices - does its own volume say it reaches?
• Ghost levels - dashed lines at the distribution's densest peaks, riding parallel to the beam, marking the prices this trend keeps returning to.
The forecast attaches only to a beam's currently-visible live element - its forming sketch, or its locked ink - and keeps no history . It is a read of the present trend, refreshed at the live edge, not a replay of the past.
🔸 THE ENGINE DIALS
• Mode - the tempo. Fast locks, breaks and re-forms sooner (short swings); Slow demands more evidence and holds through more noise (long moves); Medium is the balanced reference.
• Precision - how much data the engine reads: the perceptual fit window and the intrabar sample rate. Higher tiers resolve finer structure at more load. Sampling is timeframe-aware and never drops below one minute.
🔹 LAYERS, COLOUR & LEGIBILITY
Every layer is a toggle - forming lines, broken history, audit badges, and the forecast - so you can run it as a bare two-line channel or a fully dressed read. Colours come from three clean anchors: Support , Resistance , and Chrome (badges and neutral furniture). The whole translucent forecast - band, profile, and ghost levels - is coloured in the Oklab perceptual space, so strength shows up as vibrancy at a constant lightness rather than as glare, and a single Contrast dial scales the entire forecast from a whisper to bold.
🔸 HOW TO READ IT
• Treat a forming (dotted) line as a hypothesis and a locked (solid) line as a committed level - the tool is telling you which is which on purpose.
• Watch the audit badge : rising wick-through and falling survival probability say a locked line is wearing out.
• Read a broken line as a failed level that still marks where the structure gave way.
• Use each beam's band to see where its trend has held its volume, its reach profile for how far the trend's own volume says it can run, and its ghost levels for the prices it keeps returning to.
🔹 INPUTS
• Trend Engine - Mode (tempo) and Precision (data depth).
• Layers - show forming lines, broken history, audit badges, and the forecast.
• Colors - Support, Resistance, and Chrome anchors, plus a Contrast control for the translucent forecast.
• Channels - enable the support and/or resistance side independently.
🔸 LIMITATIONS AND HONEST NOTES
• This is a drawing and modeling assistant , not a validated strategy. It makes no performance claim and no edge claim . Nothing here is financial, investment or trading advice.
• Locked and broken lines do not repaint - once a line locks, its geometry is frozen. Forming lines move by design (they are the live sketch, and are dotted to say so), and each beam's forecast (band, profile, ghost levels) refreshes at the live edge as new volume arrives and attaches only to the current live element. These are live reads, on purpose; none of them rewrites confirmed history.
• Survival probability, maturity, remaining length and the reach profile are model estimates from the trend's own statistics - projections, not guarantees, and not forecasts of price.
• Intrabar sampling is subject to your plan's intrabar data limits ; higher Precision tiers read more intrabar data.
• Drawing budgets are finite. The tool caps its lines, labels and polylines internally, but very long histories with everything enabled push against PulseWire's per-script drawing limits - trim the layers you don't need.
Indicator
AI K-Means Clustering [TradingFinder] Machine Learning Zones🔵 Introduction
K-Means clustering is an unsupervised machine learning algorithm that groups similar data points around repeatedly updated cluster centers. Each observation is assigned to its nearest center, the centers are recalculated, and the process continues until the clusters converge. In financial market analysis, this structure can separate recurring patterns in price movement, trend direction, volume pressure, and volatility without depending entirely on fixed thresholds. As a result, the same candle may be interpreted differently in a quiet market, a directional trend, or a volatility shock, because its meaning is evaluated in relation to the surrounding market data.
This PulseWire indicator applies K-Means machine learning through several connected analysis modules. The Market State engine studies trend bias, price slope, and relative volume pressure to classify the current market regime as an active bullish trend, active bearish trend, soft bullish trend, soft bearish trend, neutral range, or low-volume range. It also compares the current cluster with the dominant cluster across recent candles, helping the trend classification remain more stable when a single large candle, temporary spike, or short-lived price reversal appears.
The Price Zones engine clusters pivot points, historical highs, and historical lows to create dynamic K-Means support and resistance zones. Traders can display all price cluster centers, the nearest K-Means zone, or separate support and resistance lines. Raw, Smooth, and Locked Steps modes control how quickly the zones respond to new price data, while the nearest line changes color according to the detected bullish, bearish, or ranging market state. A Stochastic moving average heatmap is also plotted between the outer zones, adding a visual layer for momentum, overbought and oversold conditions, trend strength, and changing market pressure.
The indicator also combines volatility analysis, price action recognition, cluster quality scoring, and alert conditions. The volatility engine uses normalized ATR, candle range, and return volatility to identify low-volatility compression, normal volatility, high volatility, and volatility shock. The Price Action module evaluates the latest closed candle for bullish and bearish zone breakouts, rejection patterns, momentum candles, and indecision near a clustered price level. A dedicated Quality and Reliability section then measures zone strength, cluster fit, zone width, price distance, and RMSE, helping traders understand whether the current machine learning calculations are strong enough for practical analysis or should be treated only as additional market context.
🔵 How to Use
The easiest way to read this indicator is not to search for one isolated green or red message. Its main value comes from combining several layers of market information: K-Means market state classification, adaptive price zones, price action, volatility conditions, and calculation quality. Each module answers a different question, and the strongest setups usually appear when several modules point in the same direction.
Start with the Market State row in the analysis table. This module applies multidimensional K-Means clustering to trend bias, trend slope, and relative volume pressure. The current cluster shows where the latest market data has been assigned, while the dominant cluster represents the most frequent cluster across the selected state window. The Strength value shows how dominant that cluster is within the recent sample.
The Market State analysis can return the following conditions :
Active Bullish Trend : Positive trend structure supported by stronger relative volume.
Soft Bullish Trend : Positive directional structure, but with weaker participation or less convincing momentum.
Active Bearish Trend : Negative trend structure supported by stronger relative volume.
Soft Bearish Trend : Bearish directional structure that still requires confirmation.
Neutral Range : Trend bias and slope are not strong enough to define a clear direction.
Low-Volume Range : Sideways structure accompanied by relatively weak volume participation.
The distinction between the current and dominant cluster is important. A single large candle can move the current data point into another cluster, but the dominant state may remain unchanged if the broader recent structure still belongs to the previous market regime. This can help prevent every temporary spike, pullback, or abnormal candle from being interpreted as a complete trend reversal.
The next section is Price Zones. Here, K-Means clustering is applied to historical pivot levels, sampled highs, and sampled lows. Instead of drawing a level from only one swing point, the algorithm groups similar historical prices and calculates a center for each price cluster. These cluster centers become adaptive K-Means price zones that may act as support, resistance, breakout references, or reaction areas.
The table displays :
Near : The cluster currently closest to price.
Strength : The percentage of sampled price levels assigned to the nearest cluster.
Nearest : The closest stabilized K-Means zone.
Support : The nearest valid cluster center below the market.
Resistance : The nearest valid cluster center above the market.
A higher Zone Strength means a larger share of the sampled levels belongs to that cluster. However, this should not be interpreted as a guaranteed support or resistance level. It simply shows that more historical observations were grouped around the same price area.
On the chart, users can choose between three visual approaches. Show All K-Means Zone Centers plots the complete set of clustered price levels. Show Nearest Zone displays only the closest stabilized level, while Show K-Means Support/Resistance plots the nearest support and resistance separately.
The nearest line changes color with the detected market state :
Green indicates a bullish market state.
Red indicates a bearish market state.
Blue indicates a neutral or ranging market state.
The zone lines can also be displayed in Raw, Smooth, or Locked Steps mode. Raw mode follows newly calculated cluster centers directly. Smooth mode gradually moves the plotted level toward the new center, creating a more stable visual structure. Locked Steps mode keeps the previous level in place until the new cluster center has moved by a meaningful ATR-based distance.
Between the outer K-Means zones, the indicator draws a Stochastic Moving Average Heatmap. This heatmap is based on a 100-period Stochastic value smoothed with a 50-period exponential moving average. Lower smoothed Stochastic values appear toward the blue and purple side of the color range, middle values move through cyan and green, and higher values progress toward yellow, orange, and red. The heatmap should be read as a visual momentum layer rather than as a standalone buy or sell signal.
The Price Action row studies candle structure in relation to the nearest K-Means zone and recent price behavior. It uses the candle body, upper wick, lower wick, previous high, previous low, and the location of the nearest zone to identify several possible conditions:
Bullish or bearish zone breakout.
Bullish or bearish rejection from a zone.
Bullish or bearish momentum candle.
Indecision at a K-Means zone.
General indecision.
No clear price action.
The Body, Upper Wick Ratio, and Lower Wick Ratio values represent the relative size of the candle body, upper wick, and lower wick compared with the candle’s total range. These values help explain why the indicator classified a candle as momentum, rejection, or indecision. Price Action should always be read together with Market State and Volatility. For example, a bullish momentum candle inside a bearish market state does not automatically create a bullish setup.
The Volatility module runs a separate K-Means model using normalized ATR, candle range percentage, and return volatility. The clustered volatility data is then used to identify four practical market conditions:
Low Volatility Compression : Market movement has contracted and a future expansion may develop;
Normal Volatility : Current movement is close to its recent reference level;
High Volatility : Price movement is elevated and may require smaller position size or wider risk parameters;
Volatility Shock : Abnormal expansion is present, making immediate entries more sensitive to slippage, unstable movement, and rapid reversals.
Volatility acts as a risk filter for the rest of the analysis. Even when Market State and Price Action point in the same direction, a High Volatility or Volatility Shock reading should reduce the confidence placed on an immediate entry.
Finally, review the Quality row. This section provides an internal assessment of how compact, representative, and consistent the current K-Means calculations are. It does not measure future profitability or win rate. Instead, it evaluates the statistical structure of the active price clusters.
The main values include :
Price Q : A combined score based on zone strength, width, fit, and price distance;
Trust : A weighted score combining price-zone quality, market-state dominance, and volatility-cluster dominance;
Fit RMSE : The normalized root mean squared error of the price clusters;
Width : The average dispersion of the nearest cluster around its center;
Reliability : A descriptive grade derived from the internal Trust score.
A narrow cluster with reasonable strength and lower fitting error will usually receive a better score than a wide, weak, or poorly fitted cluster. Use this section to decide how much weight should be given to the current analysis. A weak Quality score does not make the chart unusable, but it suggests that the levels and classifications should be treated as secondary context.
🟣 Bullish Market Reading
A bullish setup becomes more meaningful when the market state, K-Means zones, candle behavior, volatility, and quality readings support the same interpretation.
Check the Market State first : An Active Bullish Trend indicates stronger bullish structure and relative participation. A Soft Bullish Trend still favors the upside, but entries should normally wait for additional confirmation.
Locate price relative to the nearest zone : When price is above the nearest K-Means zone, that level may become an adaptive support reference. A pullback toward the green nearest-zone line can be watched for continuation or rejection behavior.
Look for bullish price action : A Bullish Rejection From Zone suggests that price tested a clustered level and closed with a stronger lower-wick reaction. A Bullish Zone Breakout shows that the candle crossed above the zone with a sufficiently large body. A Bullish Momentum Candle confirms upward pressure, but it is more useful when the Market State is already bullish.
Use the support line as a reference, not an automatic entry : The K-Means support level can help define the area where bullish structure remains valid. A decisive move below it may weaken the long scenario, especially if the Market State also changes.
Confirm volatility conditions : Normal Volatility is generally easier to manage than High Volatility or Volatility Shock. During compression, traders may wait for a confirmed breakout rather than entering before expansion begins.
Review Quality and Reliability : Stronger Quality, Trust, and Zone Strength readings increase the internal consistency of the analysis. Weak scores suggest that the zone may be broad, poorly fitted, or based on a less concentrated cluster.
A practical bullish sequence may therefore look like this: the table shows a Soft or Active Bullish Trend, price remains above or retests a green K-Means zone, a bullish rejection or breakout appears, volatility is not classified as a shock, and Quality remains acceptable. None of these elements guarantees continuation, but their alignment creates a clearer bullish context than any single reading alone.
🟣 Bearish Market Reading
Bearish analysis follows the same process in reverse. The objective is to identify whether downward market structure, clustered resistance, candle behavior, and volatility are supporting the same scenario.
Begin with the Market State : An Active Bearish Trend represents stronger negative bias, slope, and relative volume pressure. A Soft Bearish Trend favors short-side analysis but still requires confirmation before treating the move as established.
Observe price relative to the nearest zone : When price is below the nearest K-Means zone, that level may act as an adaptive resistance reference. A return toward the red nearest-zone line can be monitored for rejection or continuation.
Wait for bearish price action : A Bearish Rejection From Zone appears when price tests a clustered area and forms a stronger upper-wick reaction. A Bearish Zone Breakout indicates that price has crossed below the zone with a sufficiently large bearish body. A Bearish Momentum Candle carries more weight when the broader Market State is already bearish.
Use the resistance line to define context : The K-Means resistance level can help identify where bearish continuation remains structurally reasonable. A sustained break above it may weaken the short scenario, particularly if Market State also shifts toward bullish or neutral conditions.
Do not ignore volatility warnings : A bearish candle during Volatility Shock may be followed by a sharp continuation, but it can also produce rapid retracement and unstable execution. In this condition, the indicator explicitly favors additional confirmation or reduced risk.
Check cluster quality before relying on the level : A weak or wide price cluster may produce a less precise resistance reference. Higher Quality and Reliability readings indicate a more compact and internally consistent zone, not a guaranteed bearish outcome.
A clearer bearish sequence may include a Soft or Active Bearish Trend, price trading below or retesting a red K-Means zone, bearish rejection or breakout behavior, manageable volatility, and an acceptable Quality score. When these components disagree, for example, a bullish momentum candle inside a bearish trend, the table should be read as a warning that momentum alone is not enough to confirm a reversal.
The built-in alert conditions can be used to monitor bullish and bearish K-Means zone breakouts and rejections. Alerts are most useful as notifications that a specific price-action condition has appeared; the final interpretation should still include Market State, Volatility, zone position, and Quality before any trading decision is made.
🔵 Settings
🟣 K-Means Engine Settings
Market State Lookback : Number of recent bars used to cluster trend bias, slope, and relative volume for market-state classification.
Price Zone Lookback : Number of recent bars used to build K-Means price zones from pivots, highs, and lows.
Volatility Lookback : Number of recent bars used to cluster ATR percentage, candle range, and return volatility.
Market State Clusters : Number of clusters used by the Market State model.
Price Zone Clusters : Number of price clusters used to calculate adaptive zone centers.
Volatility Clusters : Number of clusters used by the Volatility model.
Max K-Means Iterations : Maximum number of center-update cycles allowed during each clustering calculation.
Dominant State Window : Number of recent cluster assignments used to determine the dominant market state.
Fast Volatility State Window : Number of recent volatility assignments used to determine the dominant short-term volatility cluster.
Convergence Tolerance : Minimum center movement required to continue the K-Means iteration; lower values increase precision but may require more processing.
🟣 Price Zone Settings
Pivot Length : Number of bars used on each side of a candle to confirm pivot highs and pivot lows.
High/Low Sampling Step : Controls how frequently historical highs and lows are added to the price-zone dataset; lower values use more samples.
Minimum Near-Zone Distance (%) : Minimum percentage distance used to classify price as testing a K-Means zone.
🟣 Execution Control Settings
Historical Calculation Bars : Number of recent historical bars on which calculations and visual outputs are processed.
Refresh Every N Bars : Runs the main K-Means modules once every selected number of bars and always updates them on the latest bar.
🟣 Zone Stabilizer Settings
Zone Plot Mode : Selects how zone lines are displayed: Raw follows new centers directly, Smooth moves gradually, and Locked Steps updates only after a meaningful price shift.
Zone Smooth Length : Controls the smoothing speed in Smooth mode; higher values produce slower and more stable zone movement.
Zone Lock ATR Multiplier : Defines the minimum ATR-based movement required before a zone updates in Locked Steps mode.
Nearest Zone Switch Margin ATR : Prevents frequent switching between nearby zones by requiring the new zone to be closer by an ATR-based margin.
🟣 Display Settings
Show Analysis Table : Shows or hides the market analysis table.
Table Text Size : Sets the size used inside the table.
Table Position : Selects the table location on the chart.
Show All K-Means Zone Centers : Displays all calculated K-Means price-zone centers.
Show Nearest Zone : Displays the stabilized zone closest to the current price, colored by the detected market state.
Show K-Means Support/Resistance : Displays the nearest clustered support below price and resistance above price.
🔵 Conclusion
This indicator brings K-Means clustering, market state analysis, adaptive price zones, volatility classification, and price action context into one structured workflow. Instead of reducing the chart to a single signal, it separates the market into several readable layers: directional behavior, clustered support and resistance areas, candle reactions, volatility conditions, and the internal quality of the current calculations. This makes it easier to understand whether price is trending, ranging, testing a K-Means zone, reacting to a clustered level, or moving through an unstable volatility phase.
Its strongest use comes from confirmation rather than prediction. A bullish or bearish reading becomes more meaningful when the Market State, nearest K-Means zone, Price Action module, Volatility analysis, and Quality score support the same scenario. When these components disagree, the table highlights that uncertainty instead of hiding it. Used this way, the tool works as a machine learning market analysis framework that helps organize recent price data, compare changing market regimes, and identify areas where further confirmation is still required. Indicator
NLMS Adaptive Trend Filter [BackQuant]NLMS Adaptive Trend Filter
Overview
The NLMS Adaptive Trend Filter is a machine learning inspired trend-following indicator built around one of the most important adaptive filtering algorithms in signal processing: the Normalized Least Mean Squares (NLMS) filter .
Unlike traditional moving averages that use fixed weighting schemes, the NLMS filter continuously learns from incoming market data and updates its internal coefficients in real time. Rather than assuming that price behavior remains constant, the filter attempts to adapt its structure as market conditions evolve.
This approach originates from the field of digital signal processing, where adaptive filters have been used for decades in applications such as:
• Telecommunications
• Radar systems
• Echo cancellation
• Noise reduction
• Speech processing
• Control systems
• Financial signal extraction
The goal of this indicator is to bring one of these adaptive filtering concepts into market analysis by creating a trend model that continually adjusts itself based on prediction error rather than relying on static averaging methods.
Historical Background
The roots of the NLMS filter can be traced back to the work of Bernard Widrow and Ted Hoff in the late 1950s and early 1960s.
While working at Stanford University, they developed what became known as the:
Least Mean Squares (LMS) Algorithm
The LMS algorithm was revolutionary because it provided a computationally simple method for training adaptive systems using gradient descent.
Rather than solving a complex optimization problem all at once, the LMS algorithm updates its weights incrementally after each observation.
The basic concept was:
1. Make a prediction.
2. Measure the prediction error.
3. Adjust the model slightly.
4. Repeat indefinitely.
This idea eventually became one of the foundational concepts behind modern machine learning and online optimization.
Many modern neural networks still rely on the same underlying principle:
Error → Gradient → Weight Update
The LMS algorithm later evolved into several variants, one of the most important being:
Normalized Least Mean Squares (NLMS)
NLMS improves stability by scaling weight updates according to the energy of the input signal.
This prevents learning rates from becoming too aggressive during high-volatility periods and too weak during low-volatility periods.
As a result, NLMS became one of the most widely used adaptive filtering algorithms in engineering.
What Makes NLMS Different From Moving Averages?
Traditional moving averages use predetermined weights.
For example:
Simple Moving Average (SMA)
Every observation receives equal weight.
Example:
20-period SMA
Each bar contributes:
1 / 20 = 5%
regardless of market conditions.
Exponential Moving Average (EMA)
Recent observations receive more weight.
The weighting structure is fixed and never changes.
Weighted Moving Average (WMA)
Uses linearly decreasing weights.
Again, the weighting scheme is fixed.
The problem is that markets do not operate under fixed conditions.
Volatility changes.
Trend persistence changes.
Noise levels change.
Market structure changes.
Yet traditional moving averages continue using the exact same weighting model.
NLMS takes a different approach.
Instead of assigning permanent weights, it learns them dynamically.
The filter constantly asks
"What weighting structure would have predicted the current market best?"
It then updates itself accordingly.
The Core Idea Behind Adaptive Filters
Imagine trying to forecast today's price using the previous 20 bars.
A normal moving average assumes a fixed weighting pattern.
An adaptive filter attempts to learn the optimal weighting pattern.
At every bar:
• A prediction is generated.
• Actual price is observed.
• Prediction error is measured.
• Weights are adjusted.
The process repeats indefinitely.
Over time, the filter learns which historical observations are most useful and which are less important.
Understanding Filter Taps
One of the most important concepts in adaptive filtering is the idea of:
Taps
A tap is simply a historical observation used as an input.
If the indicator uses:
20 taps
it means:
Price
Price
Price
...
Price
are all being used to generate the prediction.
Each tap receives a learned weight.
Instead of:
Current Estimate =Average of past 20 bars
the filter becomes:
Current Estimate =
(w1 × Price ) +
(w2 × Price ) +
(w3 × Price )
...
(w20 × Price )
The weights are continuously adjusted through learning.
How Prediction Works
The indicator attempts to estimate current price using previous observations.
Mathematically:
Prediction = Σ(weight × historical price)
This prediction becomes the filter output.
If the prediction is accurate:
Weights change very little.
If the prediction is poor:
Weights adjust more aggressively.
This allows the model to gradually adapt to changing market conditions.
Prediction Error
The engine measures:
Error = Actual Price − Predicted Price
This error drives all learning.
Large error means:
The model is wrong.
Small error means:
The model is performing well.
The objective is to minimize prediction error over time.
The LMS Learning Rule
The original LMS update rule is:
New Weight =Old Weight + Learning Rate × Error × Input
This is effectively a form of gradient descent.
The filter moves its weights in the direction that reduces future prediction error.
This is conceptually identical to many machine learning optimization methods.
Why Normalization Matters
The original LMS algorithm has a weakness.
When input values become very large:
Weight updates can become unstable.
This is particularly problematic in financial markets where volatility constantly changes.
NLMS solves this problem by normalizing updates according to signal energy.
Instead of:
Weight Update ∝ Error
it becomes:
Weight Update ∝ Error / Signal Power
This creates adaptive scaling.
When volatility expands:
Updates automatically shrink.
When volatility contracts:
Updates automatically expand.
This improves stability significantly.
How the Indicator Uses NLMS
The script implements an online one-step predictor.
For every new bar:
1. Previous M bars are gathered.
2. Current price is predicted.
3. Prediction error is calculated.
4. Weight vector is updated.
5. New estimate becomes available.
This process occurs continuously as new data arrives.
Because no future data is used, the filter remains fully causal and suitable for live trading.
Weight Initialization
Initially all weights are equal:
1 / M
This effectively starts the model as a simple moving average.
Over time the filter learns a custom weighting structure based on market behavior.
The initial equal-weight state acts as a neutral prior.
Step Size (μ)
The learning rate controls how aggressively the filter adapts.
Lower values:
• More stable
• Smoother output
• Slower adaptation
Higher values:
• Faster adaptation
• More responsiveness
• Greater noise sensitivity
Think of μ as controlling the intelligence speed of the model.
Small values make it conservative.
Large values make it reactive.
Regularization (ε)
Regularization prevents division by very small values.
Without it:
Periods of extremely low signal power could create unstable updates.
Regularization improves numerical stability and robustness.
It acts as a safety mechanism for the learning process.
Output Smoothing
After the NLMS estimate is generated, an optional EMA can be applied.
This smoothing is not part of the NLMS algorithm itself.
It exists purely for visual clarity.
The raw adaptive filter already contains the learning logic.
The smoothing stage simply reduces small fluctuations.
Setting smoothing to 1 effectively disables it.
Trend Detection
Trend direction is derived from the slope of the adaptive filter.
Bullish:
NLMS Output > Previous Output
Bearish:
NLMS Output < Previous Output
This creates a directional state machine.
Unlike crossover systems, trend changes occur whenever the adaptive estimate changes slope.
Bullish Flips
A bullish signal occurs when:
Trend changes from bearish to bullish.
This means the adaptive filter has transitioned from declining to rising.
Bearish Flips
A bearish signal occurs when:
Trend changes from bullish to bearish.
This means the adaptive filter has transitioned from rising to falling.
Visual Components
The indicator includes several visualization layers.
Adaptive Filter Line
The main output of the NLMS model.
This represents the learned trend estimate.
Gradient Fill
The space between price and filter is colorized.
Price Above Filter:
Bullish shading.
Price Below Filter:
Bearish shading.
This provides immediate visual context regarding trend alignment.
Edge Glow
An ATR-based glow surrounds price.
This helps emphasize directional conditions while improving chart readability.
Trend Candles
Candles can optionally inherit trend coloration.
Green:
Adaptive trend rising.
Red:
Adaptive trend falling.
This allows traders to visualize the model's directional state directly on price.
How It Differs From Traditional Trend Filters
Most trend indicators answer:
"What is the average price?"
NLMS attempts to answer:
"What weighting structure best predicts current price?"
This distinction is extremely important.
The indicator is not simply smoothing price.
It is continuously learning how price behaves.
Traditional indicators use fixed mathematics.
NLMS uses adaptive mathematics.
Strengths
• Self-adjusting weighting structure.
• Adapts to changing market conditions.
• Based on established signal-processing theory.
• Stable due to normalization.
• Less reliant on arbitrary moving-average formulas.
• Learns continuously.
• Fully causal and non-lookahead.
Limitations
• Not a predictive model in the forecasting sense.
• Can still lag during major regime shifts.
• Excessively large learning rates may introduce noise.
• Small tap counts can become unstable.
• Large tap counts can become sluggish.
Like all adaptive systems, there is a tradeoff between responsiveness and stability.
Best Use Cases
The NLMS Adaptive Trend Filter is particularly effective for:
• Trend identification.
• Regime classification.
• Dynamic support/resistance visualization.
• Adaptive trend following.
• Noise reduction.
• Signal confirmation.
Summary
The NLMS Adaptive Trend Filter applies one of the most important adaptive algorithms in modern signal processing to financial markets. Rather than relying on fixed moving-average weights, it continuously learns from prediction error and updates its internal model in real time. Built upon the pioneering work of Widrow and Hoff, the indicator combines adaptive filtering, normalized gradient descent, and online learning principles into a practical trend-following tool that evolves alongside changing market conditions. The result is a trend model that is fundamentally different from traditional moving averages, not because it smooths price differently, but because it learns how to smooth price as new information arrives.
Indicator
Neural Weight Oscillator (Zeiierman)█ Overview
The Neural Weight Oscillator (Zeiierman) is an adaptive multi-factor oscillator that combines structured decision-making with dynamic market learning.
The script analyzes three core market behaviors: Trend, Mean Reversion, and Momentum. Instead of treating these components equally, the oscillator uses the Best-Worst Method (BWM) to determine which market behavior should have the greatest influence under current market conditions.
An adaptive training layer then studies historical market reactions and gradually amplifies the features that have recently produced the strongest directional behavior.
The result is a hybrid oscillator that blends:
Human-defined market logic
Adaptive feature weighting
Multi-factor momentum analysis
Dynamic market learning
Unlike traditional oscillators that rely on static formulas, the Neural Weight Oscillator continuously adjusts its internal structure based on both trader-defined weighting preferences and changing market behavior.
█ How It Works
⚪ Market Structure Engine
The oscillator builds its analysis from three independent behavioral models: Trend, Mean Reversion, and Momentum.
The Trend component measures structural direction by comparing the fast EMA against the slow EMA, then adds the EMA slope to capture acceleration.
trendSpread = (emaFast - emaSlow) / atr
trendSlope = (emaFast - emaFast ) / atr
trendScore = normalize(trendSpread + trendSlope, -2.5, 2.5)
The Mean Reversion component measures stretched conditions using RSI exhaustion and statistical deviation from the market mean.
zScore = dev == 0 ? 0 : (close - basis) / dev
meanScore = (100 - rsi) * 0.5 + normalize(-zScore, -2.5, 2.5) * 0.5
The Momentum component measures directional acceleration using ROC, RSI momentum, and EMA velocity.
rocNorm = normalize(close / close - 1.0, -0.05, 0.05)
momentumScore = rocNorm * 0.45 + rsi * 0.35 + emaMomentum * 0.20
Each component produces its own normalized score before being blended into the final oscillator.
⚪ Best-Worst Method (BWM)
The core weighting system in the oscillator is based on the Best-Worst Method (BWM), a structured decision-making framework that creates balanced weighting relationships among multiple factors.
bestIdx = criterionIndex(bestCriterion)
worstIdx = criterionIndex(worstCriterion)
array.set(bo, bestIdx, 1.0)
array.set(ow, worstIdx, 1.0)
Instead of assigning arbitrary percentages manually, BWM allows the trader to define which market behavior matters most and which matters least. The script then automatically calculates balanced internal weights.
The process begins by selecting:
The “Best” factor → the market behavior trusted most
The “Worst” factor → the market behavior trusted least
relWeight = math.sqrt((aBW / boVal) * owVal)
The oscillator then compares all remaining factors relative to those two extremes and converts those relationships into normalized internal weights.
⚪ How To Think About The BWM Weights
The easiest way to think about BWM is:
“What type of market behavior do I trust most in the current environment?”
Different market conditions naturally favor different behaviors.
In strong directional trends , traders often prioritize Trend because structural continuation becomes the dominant force.
In choppy or range-bound markets , Mean Reversion may become more important because the market repeatedly returns back toward equilibrium.
During aggressive breakout environments , Momentum may deserve the highest weighting because acceleration becomes the primary driver.
The goal is not to find a “perfect” weight configuration, but rather to align the oscillator with the type of behavior currently dominating the market.
⚪ Adaptive Neural Training Layer
The oscillator includes an adaptive learning layer that learns how the market has recently reacted to the model’s internal features.
The script looks back at prior Trend, Mean Reversion, and Momentum feature values, then compares them to the future price reaction.
target = close / close - 1.0
targetDirection = target > 0 ? 1.0 : target < 0 ? -1.0 : 0.0
High-quality samples are ranked by how strong the move was relative to volatility.
sampleScore = math.abs(target) / qualityVol
The model then compares its internal prediction against the actual market direction and adjusts the learned feature weights over time.
pred = twTrend * s.trend + twMean * s.mean + twMomentum * s.momentum + tbias
err = pred - s.target
This allows the oscillator to gradually learn which features are producing the strongest directional behavior.
⚪ Adaptive Feature Amplification
The learned weights are converted into feature amplifiers.
trendAmplifier = 1.0 + learnTrend * blend
meanAmplifier = 1.0 + learnMean * blend
momentumAmplifier = 1.0 + learnMomentum * blend
This allows stronger features to gain more influence, while weaker features receive less influence.
█ How to Use
⚪ Reading the Oscillator
The oscillator operates between 0 and 100.
Values above 50 suggest bullish pressure dominates the market, while values below 50 suggest bearish pressure dominates.
As the oscillator moves farther away from the neutral 50 level, directional imbalance becomes stronger.
Readings above 70 typically indicate strong bullish expansion, while readings below 30 indicate strong bearish pressure. Extreme zones above 80 or below 20 may signal exhaustion conditions where reversals become more likely.
⚪ Using the BWM Weighting System
The BWM system allows traders to align the oscillator with current market behavior by controlling how much influence Trend, Mean Reversion, and Momentum should have inside the model.
Imagine the market is trending strongly upward.
You may believe:
Trend is the dominant market behavior.
Mean Reversion still matters during pullbacks.
Momentum should have the least influence.
In this case, you could choose:
Best = Trend
Worst = Momentum
You then control how strongly Trend dominates the other factors through the comparison inputs.
For example:
Best-to-Others:
Trend = 1
Mean = 3
Mom = 6
Relative-to-Worst:
Trend = 4
Mean = 2
Mom = 1
This tells the oscillator:
Trend is selected as the strongest market behavior.
Momentum is selected as the weakest market behavior.
Trend is 3x more important than Mean Reversion.
Trend is 6x more important than Momentum.
Mean Reversion is 2x more important than Momentum.
The script automatically converts these relationships into balanced internal weights.
As a result, the oscillator becomes more trend-sensitive while reducing the influence of short-term momentum fluctuations and weak counter-trend behavior.
If the market becomes highly rotational or range-bound, traders may instead increase the importance of Mean Reversion so the oscillator becomes more responsive to exhaustion and reversal conditions.
During aggressive breakout environments, increasing Momentum weighting can help the oscillator react faster to acceleration phases.
The weighting system is designed to adapt the oscillator’s personality to different market environments rather than forcing one static interpretation onto every condition.
█ Settings
Fast EMA: controls the responsiveness of the Trend and Momentum calculations.
Slow EMA: controls the structural trend baseline used throughout the oscillator.
Smoothing: controls the smoothness of the final oscillator line.
The Best and Worst: determine how the BWM weighting model prioritizes market behaviors.
Best-to-Others: define how strongly the selected Best factor dominates the remaining components.
Relative-to-Worst: define how much stronger each component is compared to the selected Worst factor.
Use Training: enables the adaptive learning layer.
Influence: controls how strongly the learned model amplifies features.
Line Impact: controls how much the adaptive model can directly influence the oscillator line itself.
-----------------
Disclaimer
The content provided in my scripts, indicators, ideas, algorithms, and systems is for educational and informational purposes only. It does not constitute financial advice, investment recommendations, or a solicitation to buy or sell any financial instruments. I will not accept liability for any loss or damage, including without limitation any loss of profit, which may arise directly or indirectly from the use of or reliance on such information.
All investments involve risk, and the past performance of a security, industry, sector, market, financial product, trading strategy, backtest, or individual's trading does not guarantee future results or returns. Investors are fully responsible for any investment decisions they make. Such decisions should be based solely on an evaluation of their financial circumstances, investment objectives, risk tolerance, and liquidity needs.
Indicator
NeuraLib Expansion: Advanced Model LayersNeuraLib_Models is the companion model expansion for NeuraLib .
NeuraLib provides the runtime: tensors, graph execution, datasets, scalers, losses, optimizers, training, inference, and validation tools. NeuraLib_Models builds on that foundation with higher-level neural architectures that are difficult and repetitive to write by hand.
The purpose of this expansion is to keep the main NeuraLib runtime clean, compact, and general, while giving researchers ready-to-use model families for sequence learning, attention, temporal pattern extraction, and Reinforcement Learning workflows.
----------------------------------------------------------------------------------------------------------------
🔷 HOW IT FITS INTO NEURALIB
NeuraLib_Models is built entirely on top of the public NeuraLib API. It does not replace the main runtime and it does not introduce a separate training engine.
After importing NeuraLib_Models, its fluent methods become available directly on NeuraLib `Sequential` models. The expansion alias can remain unused in the layer chain.
//@version=6
indicator("NeuraLib Models Quick Start", overlay = false, calc_bars_count = 600)
import Alien_Algorithms/NeuraLib/1 as nl
import Alien_Algorithms/NeuraLib_Models/1 as models
var nl.Sequential model = nl.sequential("advanced_model")
var float qLong = na
var float qFlat = na
var float qShort = na
if barstate.isfirst
model := model
.input(array.from(8), "sequence")
.temporalConvStack(4, 2, 2, 2, 1, 1, nl.ActivationKind.relu, 0.0, "temporal")
.globalAvgPool1d(3, 2, "pool")
.duelingQHead(4, 3, nl.ActivationKind.relu, "dueling_head")
.build(nl.rng(7))
float ret0 = na(close ) or close == 0.0 ? 0.0 : close / close - 1.0
float ret1 = na(close ) or close == 0.0 ? 0.0 : close / close - 1.0
float ret2 = na(close ) or close == 0.0 ? 0.0 : close / close - 1.0
float ret3 = na(close ) or close == 0.0 ? 0.0 : close / close - 1.0
float atrValue = ta.atr(14)
float atr0 = close == 0.0 ? 0.0 : atrValue / close
float atr1 = close == 0.0 ? 0.0 : atrValue / close
float atr2 = close == 0.0 ? 0.0 : atrValue / close
float atr3 = close == 0.0 ? 0.0 : atrValue / close
bool ready = not na(ret3) and not na(atr3)
if ready
nl.Tensor state = nl.vector(array.from(ret3, atr3, ret2, atr2, ret1, atr1, ret0, atr0), "state_window")
nl.Tensor qValues = model.predict(state)
qLong := qValues.get1d(0)
qFlat := qValues.get1d(1)
qShort := qValues.get1d(2)
plot(qLong, "Q long", color = color.lime, linewidth = 2)
plot(qFlat, "Q flat", color = color.gray)
plot(qShort, "Q short", color = color.red, linewidth = 2)
hline(0.0, "Zero", color = color.new(color.gray, 70))
The model is still a normal NeuraLib model. You still call `.compile()`, `.trainOnBatch()`, `.predict()`, `.evaluate()`, `.getWeightsArray()`, and `.softUpdateFrom()` from the main library.
----------------------------------------------------------------------------------------------------------------
🔷 WHY THIS EXPANSION EXISTS
The main NeuraLib library is the foundation. It exposes a graph engine powerful enough to create custom architectures, but repeatedly building LSTM gates, attention projections, residual blocks, Conv1D stacks, or Transformer paths from raw graph operations would be too verbose for everyday research.
NeuraLib_Models packages those patterns into readable blocks:
Temporal models : Conv1D blocks, temporal convolution stacks, global average pooling, and global max pooling for flattened sequence inputs.
Recurrent models : LSTM and GRU blocks for compact sequence memory.
Attention models : Self-attention, multi-head self-attention, cross-attention, Transformer encoder blocks, Transformer encoder stacks, and Transformer decoder blocks.
Residual models : Residual dense blocks for deeper feedforward paths.
Reinforcement Learning heads : Q-head blocks and dueling Q-heads for action-value style outputs.
Replay utilities : Deterministic Prioritized Experience Replay for reproducible Pine research.
Sequence helpers : Positional encoding for token, sequence, and attention workflows.
----------------------------------------------------------------------------------------------------------------
🔷 PRACTICAL EXAMPLES
🔸 Temporal Conv Model With Dueling Q-Head
This pattern is useful when a flattened sequence contains recent market states and the output represents action values.
//@version=6
indicator("NeuraLib Models Temporal Q Example", overlay = false, calc_bars_count = 600)
import Alien_Algorithms/NeuraLib/1 as nl
import Alien_Algorithms/NeuraLib_Models/1 as models
var nl.Sequential qModel = nl.sequential("temporal_q_model")
var nl.WindowDataset qDataset = nl.windowDataset(8, 3, 400, "q_rows")
var float qDown = na
var float qNeutral = na
var float qUp = na
var float qLoss = na
if barstate.isfirst
nl.CompileConfig cfg = nl.compileConfig()
cfg := cfg
.presetQValues()
.optimizer(nl.adamW(0.001))
.withTrainingGate(true)
qModel := qModel
.input(array.from(8), "state_window")
.temporalConvStack(4, 2, 2, 2, 1, 1, nl.ActivationKind.relu, 0.0, "temporal")
.globalAvgPool1d(3, 2, "pool")
.duelingQHead(4, 3, nl.ActivationKind.relu, "dueling_head")
.compile(cfg)
qDataset := qDataset
.setInputScaler(nl.ScalerKind.zScore)
.setTargetScaler(nl.ScalerKind.none)
float ret0 = na(close ) or close == 0.0 ? 0.0 : close / close - 1.0
float ret1 = na(close ) or close == 0.0 ? 0.0 : close / close - 1.0
float ret2 = na(close ) or close == 0.0 ? 0.0 : close / close - 1.0
float ret3 = na(close ) or close == 0.0 ? 0.0 : close / close - 1.0
float ret4 = na(close ) or close == 0.0 ? 0.0 : close / close - 1.0
float atrValue = ta.atr(14)
float atr0 = close == 0.0 ? 0.0 : atrValue / close
float atr1 = close == 0.0 ? 0.0 : atrValue / close
float atr2 = close == 0.0 ? 0.0 : atrValue / close
float atr3 = close == 0.0 ? 0.0 : atrValue / close
float atr4 = close == 0.0 ? 0.0 : atrValue / close
bool rowReady = not na(ret4) and not na(atr4)
if rowReady
array features = array.from(ret4, atr4, ret3, atr3, ret2, atr2, ret1, atr1)
float downTarget = math.max(-ret0, 0.0)
float neutralTarget = math.max(0.002 - math.abs(ret0), 0.0)
float upTarget = math.max(ret0, 0.0)
qDataset := qDataset.pushRow(features, array.from(downTarget, neutralTarget, upTarget))
if qDataset.ready(48)
if barstate.islastconfirmedhistory
nl.Batch train = qDataset.trainBatch(12)
qModel := qModel.trainOnBatch(train.inputTensor, train.targetTensor)
qLoss := qModel.trainStats.lastLoss
nl.Tensor liveState = nl.vector(array.from(ret3, atr3, ret2, atr2, ret1, atr1, ret0, atr0), "live_state")
nl.Tensor scaledState = qDataset.scaleInput(liveState)
nl.Tensor qValues = qModel.predict(scaledState)
qDown := qValues.get1d(0)
qNeutral := qValues.get1d(1)
qUp := qValues.get1d(2)
plot(qDown, "Q down", color = color.red, linewidth = 2)
plot(qNeutral, "Q neutral", color = color.gray)
plot(qUp, "Q up", color = color.lime, linewidth = 2)
plot(qLoss, "Training loss", color = color.orange)
hline(0.0, "Zero", color = color.new(color.gray, 70))
Input shape `array.from(8)` represents a flattened 4 step by 2 feature sequence. The temporal stack extracts short sequence structure, pooling compresses the sequence, and the dueling head separates value and advantage paths before producing action scores. The example trains only on the last confirmed historical bar so it remains safe to paste onto long charts.
🔸 Transformer Encoder For Token Rows
Attention models are useful when each row is a token or time step, and each column is a feature dimension.
//@version=6
indicator("NeuraLib Models Transformer Encoder Example", overlay = false, calc_bars_count = 600)
import Alien_Algorithms/NeuraLib/1 as nl
import Alien_Algorithms/NeuraLib_Models/1 as models
var nl.Sequential encoder = nl.sequential("encoder_model")
var float tokenSignal = na
var float tokenContext = na
var float tokenVolatility = na
if barstate.isfirst
encoder := encoder
.input(array.from(4), "tokens")
.multiHeadSelfAttention(4, 2, true, "mha")
.transformerEncoder(4, true, 2, nl.ActivationKind.geluApprox, "encoder", 0.05, 2)
.build(nl.rng(11))
float emaValue = ta.ema(close, 21)
float atrValue = ta.atr(14)
float ret0 = na(close ) or close == 0.0 ? 0.0 : close / close - 1.0
float ret1 = na(close ) or close == 0.0 ? 0.0 : close / close - 1.0
float ret2 = na(close ) or close == 0.0 ? 0.0 : close / close - 1.0
float ret3 = na(close ) or close == 0.0 ? 0.0 : close / close - 1.0
float emaGap0 = emaValue == 0.0 ? 0.0 : close / emaValue - 1.0
float emaGap1 = emaValue == 0.0 ? 0.0 : close / emaValue - 1.0
float emaGap2 = emaValue == 0.0 ? 0.0 : close / emaValue - 1.0
float emaGap3 = emaValue == 0.0 ? 0.0 : close / emaValue - 1.0
float atr0 = close == 0.0 ? 0.0 : atrValue / close
float atr1 = close == 0.0 ? 0.0 : atrValue / close
float atr2 = close == 0.0 ? 0.0 : atrValue / close
float atr3 = close == 0.0 ? 0.0 : atrValue / close
bool ready = not na(ret3) and not na(emaGap3) and not na(atr3)
if ready
nl.Tensor tokens = nl.vector(array.from(
ret3, emaGap3, atr3, -1.0,
ret2, emaGap2, atr2, -0.33,
ret1, emaGap1, atr1, 0.33,
ret0, emaGap0, atr0, 1.0), "tokens").reshape(array.from(4, 4))
nl.Tensor encoded = encoder.predict(tokens)
tokenSignal := encoded.get1d(12)
tokenContext := encoded.get1d(13)
tokenVolatility := encoded.get1d(14)
plot(tokenSignal, "Latest token signal", color = color.aqua, linewidth = 2)
plot(tokenContext, "Latest token context", color = color.purple)
plot(tokenVolatility, "Latest token volatility", color = color.orange)
hline(0.0, "Zero", color = color.new(color.gray, 70))
In this example, each input row has 4 features. `headCount` is 2, so the model dimension is split into two attention heads.
Attention rule: `modelDim` must be divisible by `headCount`, and the current implementation supports up to 8 heads.
🔸 Prioritized Experience Replay
Prioritized Experience Replay stores examples with priorities, then returns reproducible weighted samples. This is especially useful for Reinforcement Learning experiments where high-error transitions should be revisited more often.
//@version=6
indicator("NeuraLib Models PER Example", overlay = false, calc_bars_count = 1200)
import Alien_Algorithms/NeuraLib/1 as nl
import Alien_Algorithms/NeuraLib_Models/1 as models
var models.PrioritizedReplayBuffer replay = models.prioritizedReplayBuffer(4, 2, 300, "replay")
var nl.Sequential replayModel = nl.sequential("replay_q_model")
var float replayLoss = na
var float firstImportanceWeight = na
var float replayRows = na
if barstate.isfirst
nl.CompileConfig cfg = nl.compileConfig()
cfg := cfg
.presetQValues()
.optimizer(nl.adamW(0.001))
.trainEveryCall()
replayModel := replayModel
.input(array.from(4), "state")
.dense(8, nl.ActivationKind.relu, "hidden")
.qHead(2, nl.ActivationKind.linear, "q_values")
.compile(cfg)
float rsiValue = ta.rsi(close, 14)
float emaValue = ta.ema(close, 21)
float atrValue = ta.atr(14)
float atrPct = close == 0.0 ? 0.0 : atrValue / close
float momentum = na(close ) or close == 0.0 ? 0.0 : close / close - 1.0
float nextReturn = na(close ) ? 0.0 : nl.nextReturnValue(close , close)
bool rowReady = not na(rsiValue ) and not na(emaValue ) and not na(atrPct ) and not na(momentum )
if rowReady
float prevEma = emaValue
float priceVsEma = prevEma == 0.0 ? 0.0 : close / prevEma - 1.0
array stateFeatures = array.from(rsiValue / 100.0, priceVsEma, atrPct , momentum )
array targetValues = array.from(math.max(-nextReturn, 0.0), math.max(nextReturn, 0.0))
float priority = math.abs(nextReturn) + 0.0001
replay := replay.pushExperience(stateFeatures, targetValues, priority)
replayRows := float(replay.size())
if replay.ready(32)
models.PrioritizedReplaySample sample = replay.sampleBatch(32, 0.6, 0.4, 17)
replayModel := replayModel.trainOnBatch(sample.batch.inputTensor, sample.batch.targetTensor)
replayLoss := replayModel.trainStats.lastLoss
firstImportanceWeight := sample.weightArray.size() > 0 ? sample.weightArray.get(0) : na
if sample.indexArray.size() > 0
replay := replay.updatePriority(sample.indexArray.get(0), replayLoss + 0.0001)
plot(replayLoss, "Replay training loss", color = color.orange, linewidth = 2)
plot(firstImportanceWeight, "First sample weight", color = color.aqua)
The returned sample includes:
batch : A normal NeuraLib `Batch` containing sampled inputs and targets.
indexArray : Logical replay indices that can be passed back to `updatePriority()`.
weightArray : Normalized importance weights for custom loss weighting or diagnostics.
sampleRows : Number of sampled rows.
PER sampling is deterministic for a given buffer, `batchSize`, and `seed`. That makes Pine tests and live research easier to reproduce.
----------------------------------------------------------------------------------------------------------------
🔷 MODEL FAMILIES
🔸 Residual Dense Blocks
`residualDense()` adds a feedforward residual block. Residual paths help preserve information through deeper models and reduce the chance that a dense stack destroys useful features too early.
🔸 Conv1D And Temporal Convolution Stacks
`conv1d()` and `temporalConvStack()` operate on flattened sequence inputs. A sequence with `timeSteps = 4` and `featureCount = 2` is represented as 8 input features. These blocks are useful for local temporal structure, short rolling windows, feature rhythm, and compact pattern extraction.
🔸 Global Pooling
`globalAvgPool1d()` and `globalMaxPool1d()` compress flattened sequence outputs into feature-level summaries. Average pooling captures broad sequence behavior, while max pooling emphasizes the strongest activation per feature.
🔸 LSTM And GRU Blocks
`lstm()` and `gru()` provide recurrent sequence memory over flattened time-series inputs. They are useful when the order of recent states matters more than a single snapshot.
🔸 Attention And Transformers
`selfAttention()`, `multiHeadSelfAttention()`, `crossAttention()`, `transformerEncoder()`, `transformerEncoderStack()`, and `transformerDecoder()` bring attention-style modeling into Pine. They are designed for compact token matrices, packed target-memory layouts, and small Transformer-style research models that fit PulseWire limits.
🔸 Q-Heads And Dueling Q-Heads
`qHeadBlock()` creates action-value style outputs. `duelingQHead()` splits the model into value and advantage branches, then recombines them into Q-values. This is useful when you want the model to estimate both the overall state value and the relative value of each action.
🔸 Positional Encoding
`pushPositionalEncoding()` adds sinusoidal position features to a NeuraLib `FeatureBuilder`. This helps attention-style models distinguish where a token or time step sits in a sequence.
----------------------------------------------------------------------------------------------------------------
🔷 FEATURE QUICK REFERENCE
Built on NeuraLib : Uses the main NeuraLib graph, tensor, training, optimizer, dataset, and inference runtime.
Fluent API : Adds methods directly to NeuraLib `Sequential` models after import.
Block factories : Provides standalone `GraphBlock` factories for users who want lower-level composition.
Temporal modeling : Conv1D, temporal convolution stacks, and 1D pooling.
Recurrent modeling : LSTM and GRU sequence blocks.
Attention modeling : Self-attention, multi-head self-attention, cross-attention, encoders, encoder stacks, and decoders.
Reinforcement Learning support : Q-heads, dueling Q-heads, target-model soft updates through NeuraLib, and Prioritized Experience Replay.
Reproducible replay : PER sampling is deterministic for a given seed.
Shape guardrails : Advanced builders validate expected model feature counts and attention head compatibility.
----------------------------------------------------------------------------------------------------------------
🔷 IMPORTANT USAGE NOTES
Import order matters : Import `NeuraLib` first, then `NeuraLib_Models`.
The alias can be unused : The imported expansion registers methods on NeuraLib types, so `.lstm()`, `.gru()`, `.transformerEncoder()`, and similar methods can be called in the model chain.
Keep models compact : Pine Script has execution limits. Start with small hidden sizes, short sequences, and low head counts.
Control chart history : Use `calc_bars_count = 600` in `indicator()` when needed to balance available training history against model size and execution time.
Respect sequence shapes : Conv1D, temporal stacks, LSTM, and GRU methods expect flattened sequence sizes of `timeSteps * featureCount`.
Respect attention shapes : Attention methods expect each input row to have `modelDim` columns. Cross-attention and decoder blocks use packed rows.
Use NeuraLib guardrails : Train/validation splits, scalers, EarlyStopper, training gates, and gradient clipping remain part of the main NeuraLib workflow.
----------------------------------------------------------------------------------------------------------------
🔷 API REFERENCE
🔸 Sequential Methods
residualDense(hiddenUnits, activationKind, dropoutRate, name) : Adds a residual dense block.
duelingQHead(hiddenUnits, actionCount, activationKind, name) : Adds a dueling value/advantage Q-head.
conv1d(timeSteps, featureCount, filters, kernelSize, stride, activationKind, name) : Adds a Conv1D block for flattened sequences.
temporalConvStack(timeSteps, featureCount, filters, kernelSize, layers, stride, activationKind, dropoutRate, name) : Adds stacked temporal Conv1D layers.
globalAvgPool1d(timeSteps, featureCount, name) : Adds global average pooling over a flattened 1D sequence.
globalMaxPool1d(timeSteps, featureCount, name) : Adds global max pooling over a flattened 1D sequence.
lstm(timeSteps, featureCount, units, activationKind, name) : Adds an LSTM scan block.
gru(timeSteps, featureCount, units, activationKind, name) : Adds a GRU scan block.
selfAttention(modelDim, causal, name) : Adds row-wise self-attention.
multiHeadSelfAttention(modelDim, headCount, causal, name) : Adds multi-head self-attention.
crossAttention(queryRows, memoryRows, modelDim, headCount, name) : Adds packed query-memory cross-attention.
transformerEncoder(modelDim, causal, ffMultiplier, activationKind, name, dropoutRate, headCount) : Adds one Transformer encoder block.
transformerEncoderStack(modelDim, layers, causal, ffMultiplier, activationKind, dropoutRate, headCount, name) : Adds repeated Transformer encoder blocks.
transformerDecoder(targetRows, memoryRows, modelDim, headCount, ffMultiplier, activationKind, dropoutRate, name) : Adds a packed target-memory Transformer decoder.
🔸 GraphBlock Factories
qHeadBlock(inputFeatures, actionCount, activationKind, name) : Creates a Q-head block.
duelingQHeadBlock(inputFeatures, hiddenUnits, actionCount, activationKind, name) : Creates a dueling Q-head block.
residualDenseBlock(inputFeatures, hiddenUnits, activationKind, dropoutRate, name) : Creates a residual dense block.
conv1dBlock(timeSteps, featureCount, filters, kernelSize, stride, activationKind, name) : Creates a Conv1D block.
temporalConvStackBlock(timeSteps, featureCount, filters, kernelSize, layers, stride, activationKind, dropoutRate, name) : Creates a temporal convolution stack.
globalAvgPool1dBlock(timeSteps, featureCount, name) and globalMaxPool1dBlock(timeSteps, featureCount, name) : Create pooling blocks.
lstmBlock(timeSteps, featureCount, units, activationKind, name) and gruBlock(timeSteps, featureCount, units, activationKind, name) : Create recurrent blocks.
selfAttentionBlock(modelDim, causal, name) , multiHeadSelfAttentionBlock(modelDim, headCount, causal, name) , and crossAttentionBlock(queryRows, memoryRows, modelDim, headCount, name) : Create attention blocks.
transformerEncoderBlock(modelDim, causal, ffMultiplier, activationKind, name, dropoutRate, headCount) and transformerDecoderBlock(targetRows, memoryRows, modelDim, headCount, ffMultiplier, activationKind, dropoutRate, name) : Create Transformer blocks.
🔸 Prioritized Experience Replay
prioritizedReplayBuffer(featureCount, targetCount, maxRows, name) : Creates a replay buffer.
pushExperience(featureRowArray, targetRowArray, priority) : Adds or overwrites one replay row.
sampleBatch(batchSize, alpha, beta, seed) : Returns a deterministic weighted sample.
updatePriority(index, priority) : Updates a sampled row priority.
toBatch() : Returns all replay rows in chronological order.
ready(minRows) , size() , and clear() : Replay buffer utilities.
🔸 Feature Helpers
pushPositionalEncoding(position, dimensions, maxPeriod, featurePrefix) : Appends sinusoidal positional encoding values to a NeuraLib `FeatureBuilder`.
NeuraLib_Models is for Pine Script developers who want higher-level neural architecture blocks without leaving the NeuraLib runtime. It is built for compact research models inside PulseWire's execution limits, not for oversized GPU-style networks.
All the diagrams in this publication are rendered natively on PulseWire using Pine3D
----------------------------------------------------------------------------------------------------------------
This work is licensed under (CC BY-NC-SA 4.0) , meaning usage is free for non-commercial purposes given that Alien_Algorithms is credited in the description for the underlying software. For commercial use licensing, contact Alien_Algorithms
Library
NeuraLib: A Native AI and Deep Learning RuntimeNeuraLib is a tensor-based, auto-differentiating Machine Learning runtime built natively for Pine Script™.
It brings real Deep Learning mechanisms that power modern Artificial Intelligence systems into PulseWire. Instead of relying on fixed formulas, static regressions, or rigid structures, NeuraLib gives Pine developers a different tool: a compact neural runtime that can learn from the features you feed it, using the architecture you define.
This means users are no longer limited to classical methods like Linear Regression, Logistic Regression, KNN, Naive Bayes, Kalman Filters, or Markov Chains. One can build adaptive architectures perfectly suited for custom indicators, strategies, regime detection, directional prediction, price transforms, and AI-assisted signal generation.
Using NeuraLib, one can build a model, collect market data, normalize it, run predictions, train through backpropagation, track validation behavior, and update weights directly inside PulseWire.
Furthermore, it is not necessary to directly display trained variables. The process can be a part of a larger script functionality, where AI-powered decision making changes how an indicator behaves.
The goal is to make real neural network workflows usable in Pine Script without hiding the important controls, being scalable with evolving market dynamics, and abstracting away the complexity that comes with such software. The provided API is highly modular and intuitive, using chained object-oriented programming for easy readability and use. The backend is engineered with fault-tolerance in mind, providing users with sanity checks and preventing common pitfalls by default.
Think of NeuraLib as a comprehensive machine learning ecosystem, containing:
A Model Builder : Define neural networks with readable chained calls like `.input()`, `.dense()`, and `.dropout()`.
An In-Pine Training Engine : Models calculate losses, backpropagate gradients, update weights, and produce predictions directly on chart data.
Automated Data Pipelines : Built-in datasets handle feature collection, robust scaling (Z-Score, Min-Max), validation holdout splits, and time-series rolling windows.
Finance-Native Loss Functions : Beyond standard error metrics, the engine includes Directional, Quantile, Multi-Horizon Weighted, and Sharpe-style losses tailored for trading.
Practical Training Controls : Layer Normalization, AdamW weight decay, gradient clipping, gradient accumulation, and early stopping are built in to prevent overfitting.
Advanced Optimizers : Train networks using RMSProp, Adam, or AdamW, paired with learning rate schedules like Warmup Cosine and Step Decay.
For newer users, this means you can start with a simple dense model. For advanced users, the same runtime exposes graph operations, custom blocks, tensors, matrix operations, optimizers, schedules, losses, and extension hooks.
In plain terms, a model receives a row of numbers called features, compares its output against a target, measures the error with a loss function, and then adjusts its internal weights to reduce that error next time.
----------------------------------------------------------------------------------------------------------------
🔷 WHAT MAKES IT DIFFERENT
🔸 Parity-tested neural math
NeuraLib’s core operations have been tested against established Machine Learning Runtimes outside of PulseWire (Such as Keras / TensorFlow / PyTorch).
The goal was not to imitate the appearance of Machine Learning, but to reproduce the math that is proven to work. Standard forward passes, gradients, losses, and optimizer behavior were checked for 1:1 algorithmic parity, with negligible differences coming from normal floating-point behavior.
That means the matrix math, backpropagation, and gradient updates running on your chart follow the same underlying logic expected from professional Machine Learning environments.
🔸 Matrix-first computation
NeuraLib uses tensor and matrix abstractions as the foundation of the runtime. Under the hood, it supports the operations needed for neural computation, including matrix multiplication, broadcasting, activation functions, softmax, slicing, concatenation, reductions, normalization, attention scoring, convolution-style operations, and recurrent scan blocks.
🔸 Auto-differentiating graph engine
NeuraLib makes the computational graph a first-class object.
You can use high-level Sequential models, or build custom GraphBlocks from lower-level operations. Once a custom block is connected to a model, the same runtime handles the backward pass. That means your custom architecture can be trained with the same `.trainOnBatch()` workflow as standard layers.
----------------------------------------------------------------------------------------------------------------
🔷 CUSTOM GRAPHS
The Sequential API is the easiest way to start, but NeuraLib is not just a list of built-in layers.
You can create a `GraphBlock`, add operations, set an output node, and plug that block into a model. Once connected, the runtime handles the backward pass and parameter updates.
Useful graph operations include:
Matrix multiplication, transpose, add, subtract, multiply, divide, and scale.
Activation functions and softmax.
Layer Normalization and Dropout.
Causal masking, slicing, concatenation, row reduction, and column reduction.
Global average pooling and global max pooling for 1D sequences.
Attention score and attention apply operations.
Conv1D, LSTM scan, and GRU scan primitives.
This is the foundation that allows companion model libraries to add advanced AI and Machine Learning architectures without changing the main NeuraLib runtime.
----------------------------------------------------------------------------------------------------------------
🔷 BUILT-IN DATA GUARDRAILS
NeuraLib is not only a training mechanism. It also includes guardrails for cleaner research:
Invalid rows are rejected : Dataset rows must match the configured feature and target counts, and rows containing `na` values are not inserted.
Shape checks protect model calls : Forward, training, backward, and evaluation paths validate input and target shapes before running expensive graph code.
Train and validation splits are separated : `trainBatch()` and `validationBatch()` use holdout rows instead of blending all rows into one batch.
Scaler leakage is controlled : Validation batches are scaled from the training-side profile where the dataset split requires it, so validation normalization does not learn from the holdout slice.
Rolling windows respect time order : `RollingDataset` supports target offsets and wrapped ring buffers while preserving chronological reads.
These checks help reduce common data poisoning and data leakage mistakes: wrong row widths, missing values, validation contamination, target-offset leakage, and accidental overtraining across every historical bar.
----------------------------------------------------------------------------------------------------------------
🔷 A FIRST MODEL
The basic API is intentionally readable. This creates a small model with dropout, one hidden layer, Huber loss, AdamW optimization, and MAE tracking.
//@version=6
indicator("NeuraLib Basic Model", overlay = false, calc_bars_count = 600)
import Alien_Algorithms/NeuraLib/1 as nl
var nl.Sequential model = nl.sequential("basic_model")
var float modelOutput = na
if barstate.isfirst
nl.CompileConfig cfg = nl.compileConfig()
cfg := cfg
.optimizer(nl.adamW(0.001))
.loss(nl.LossKind.huber)
.metric(nl.MetricKind.mae)
.withTrainingGate(true)
model := model
.input(array.from(4), "features")
.dropout(0.15)
.dense(8, nl.ActivationKind.relu, "hidden")
.dense(1, nl.ActivationKind.linear, "output")
.compile(cfg)
float rsiValue = ta.rsi(close, 14)
float emaValue = ta.ema(close, 21)
float atrValue = ta.atr(14)
float atrPct = close == 0.0 ? 0.0 : atrValue / close
float momentum = na(close ) or close == 0.0 ? 0.0 : close / close - 1.0
bool ready = not na(rsiValue) and not na(emaValue) and not na(atrPct) and not na(momentum)
if ready
float priceVsEma = emaValue == 0.0 ? 0.0 : close / emaValue - 1.0
nl.Tensor inputTensor = nl.vector(array.from(rsiValue, priceVsEma, atrPct, momentum), "features")
nl.Tensor outputTensor = model.predict(inputTensor)
modelOutput := outputTensor.get1d(0)
plot(modelOutput, "Untrained model output", color = color.aqua, linewidth = 2)
hline(0.0, "Zero", color = color.new(color.gray, 70))
The same model can then receive scaled batches from a dataset and train with `.trainOnBatch()`. The plot in this first example is the untrained forward output, included so the block can be pasted directly into an indicator.
----------------------------------------------------------------------------------------------------------------
🔷 A PRACTICAL DATA FLOW
Machine Learning models usually fail when the data pipeline is careless. Price, volume, volatility, and oscillators often live on very different scales. NeuraLib includes dataset and scaling helpers so the common workflow stays explicit:
Build a feature row.
Build a target row.
Push the row into a dataset.
Request a training batch.
Request a validation batch when needed.
Train, evaluate, predict, and inverse-scale targets when appropriate.
//@version=6
indicator("NeuraLib Return Validation Example", overlay = false, calc_bars_count = 600)
import Alien_Algorithms/NeuraLib/1 as nl
var nl.Sequential model = nl.sequential("returns_model")
var nl.WindowDataset dataset = nl.windowDataset(4, 1, 500, "returns_dataset")
var float predictedReturn = na
var float validationLossValue = na
var float trainingLossValue = na
if barstate.isfirst
nl.CompileConfig cfg = nl.compileConfig()
cfg := cfg
.optimizer(nl.adamW(0.003))
.loss(nl.LossKind.huber)
.metric(nl.MetricKind.mae)
.trainEveryCall()
model := model
.input(array.from(4), "features")
.dense(8, nl.ActivationKind.relu, "hidden")
.dropout(0.10, "dropout")
.dense(1, nl.ActivationKind.linear, "next_return")
.compile(cfg)
dataset := dataset
.setInputScaler(nl.ScalerKind.zScore)
.setTargetScaler(nl.ScalerKind.zScore)
float rsiValue = ta.rsi(close, 14)
float emaValue = ta.ema(close, 21)
float atrValue = ta.atr(14)
float atrPct = close == 0.0 ? 0.0 : atrValue / close
float momentum = na(close ) or close == 0.0 ? 0.0 : close / close - 1.0
float realizedReturn = na(close ) ? na : nl.nextReturnValue(close , close)
bool rowReady = not na(rsiValue ) and not na(emaValue ) and not na(atrPct ) and not na(momentum ) and not na(close )
if rowReady
float prevEma = emaValue
float priceVsEma = prevEma == 0.0 ? 0.0 : close / prevEma - 1.0
array features = array.from(
rsiValue ,
priceVsEma,
atrPct ,
momentum )
array target = array.from(nl.nextReturnValue(close , close))
dataset := dataset.pushRow(features, target)
if dataset.ready(64)
nl.Batch train = dataset.trainBatch(16)
nl.Batch validation = dataset.validationBatch(16)
model := model.trainOnBatch(train.inputTensor, train.targetTensor)
trainingLossValue := model.trainStats.lastLoss
nl.LossResult validationLoss = model.evaluate(validation.inputTensor, validation.targetTensor)
validationLossValue := validationLoss.value
bool liveReady = not na(rsiValue) and not na(emaValue) and not na(atrPct) and not na(momentum)
if liveReady
float livePriceVsEma = emaValue == 0.0 ? 0.0 : close / emaValue - 1.0
array liveFeatures = array.from(rsiValue, livePriceVsEma, atrPct, momentum)
nl.Tensor liveInput = nl.vector(liveFeatures, "live_features")
nl.Tensor scaledInput = dataset.scaleInput(liveInput)
nl.Tensor scaledPrediction = model.predict(scaledInput)
nl.Tensor rawPrediction = dataset.inverseScaleTarget(scaledPrediction)
predictedReturn := rawPrediction.get1d(0)
plot(realizedReturn, "Last realized return", color = color.gray)
plot(predictedReturn, "Predicted next return", color = color.aqua, linewidth = 2)
plot(validationLossValue, "Validation loss", color = color.orange)
plot(trainingLossValue, "Training loss", color = color.new(color.blue, 35))
hline(0.0, "Zero", color = color.new(color.gray, 70))
This example trains from completed historical pairs. The feature row comes from the previous bar, and the target is the return from that previous bar to the current bar. That keeps the example easy to inspect and avoids using future information in the feature row. When pasted into an indicator, it plots the last realized return, the model's predicted next return, training loss, and validation loss.
----------------------------------------------------------------------------------------------------------------
🔷 TWO PRACTICAL EXECUTION MODES
Deep Learning in Pine requires careful execution control. NeuraLib supports two main workflows.
🔸 1. Live-edge training
Use this when you want safer execution for larger models.
The dataset can collect rows across the chart, while the expensive training step only runs on the last confirmed historical bar. This helps avoid timeouts while still allowing the model to learn from recent prepared data.
cfg := cfg.withTrainingGate(true)
Use this for:
Larger models
More features
Rolling sequence inputs
Heavier architectures
Safer live-edge updates
🔸 2. Full-history training and inference
Use this when the model is intentionally small.
The model can train and infer across historical bars, which makes it possible to create lightweight adaptive indicators, such as an AI Moving Average that learns from recent local structure instead of using a fixed smoothing formula.
cfg := cfg.trainEveryCall()
Use this for:
Tiny dense models
Small batches
Fast adaptive filters
AI-assisted moving averages
Lightweight feature transforms
For full-history workflows, start small. A shallow model with 4 to 8 hidden units and a batch size of 8 or 16 is usually a better starting point than a deep architecture.
----------------------------------------------------------------------------------------------------------------
🔷 ADVANCED MODEL EXPANSION
NeuraLib is designed to act as the foundation for larger model libraries and community-built extensions.
To demonstrate this, NeuraLib Expansion: Advanced Model Layers is built entirely on top of the public NeuraLib API and is launched in parallel on day one. The expansion library is published as NeuraLib_Models . It extends the runtime with higher-level builders for LSTMs, GRUs, temporal convolution stacks, residual dense blocks, dueling Q-heads for Reinforcement Learning, Transformer-style attention blocks, and Prioritized Experience Replay utilities.
The important part is architectural: advanced models plug into the same runtime. NeuraLib remains the foundation for tensors, graph execution, optimization, training, inference, datasets, and scaling. After importing `NeuraLib_Models`, its fluent methods become available on NeuraLib `Sequential` models, so the expansion alias does not need to be referenced directly in the layer chain.
//@version=6
indicator("NeuraLib Models Extension Demo", overlay = false, calc_bars_count = 600)
import Alien_Algorithms/NeuraLib/1 as nl
import Alien_Algorithms/NeuraLib_Models/1 as models
var nl.Sequential model = nl.sequential("advanced_demo")
if barstate.isfirst
model := model
.input(array.from(8), "sequence")
.temporalConvStack(4, 2, 3, 2, 2, 1, nl.ActivationKind.relu, 0.0, "temporal")
.globalAvgPool1d(2, 3, "pool")
.duelingQHead(4, 2, nl.ActivationKind.relu, "q_head")
.build(nl.rng(7))
----------------------------------------------------------------------------------------------------------------
🔷 FEATURE QUICK REFERENCE
Runtime : Matrix-first auto-differentiating neural graph runtime for Pine Script.
Model API : Chainable `Sequential` builder with `input`, `dense`, `dropout`, `layerNorm`, `activation`, `flatten`, `reshape`, and custom `block` support.
Training : Forward pass, loss calculation, backpropagation, gradient accumulation, optimizer steps, train stats, and history buffers.
Inference : `.predict()` for deterministic inference and `.predictMC()` for dropout-based uncertainty sampling.
Datasets : `WindowDataset` for flat rows and `RollingDataset` for time-series windows.
Scaling : None, Z-Score, Min-Max, Running Z-Score scalers, dataset input scaling, target scaling, and inverse target scaling.
Optimizers : SGD, Momentum, RMSProp, Adam, and AdamW.
Schedulers : Constant, Step Decay, Cosine Decay, and Warmup Cosine.
Activations : Linear, ReLU, Leaky ReLU, ELU, GELU Approx, Sigmoid, Tanh, Softplus, Swish, and Softmax.
Losses : MSE, MAE, Huber, LogCosh, Binary Cross Entropy, Binary Cross Entropy From Logits, Categorical Cross Entropy, Softmax Cross Entropy From Logits, Directional, Quantile, Multi-Horizon Weighted, and Sharpe.
Metrics : MAE, RMSE, Directional Accuracy, Binary Accuracy, Binary Accuracy From Logits, Categorical Accuracy, and Cosine Similarity.
Guardrails : Shape validation, invalid-row rejection, train/validation split helpers, leakage-aware scaler profiles, training gates, gradient clipping, and EarlyStopper.
Advanced expansion : Conv1D, temporal stacks, recurrent blocks, attention, Transformers, dueling Q-heads, positional encodings, and Prioritized Experience Replay.
----------------------------------------------------------------------------------------------------------------
🔷 IMPORTANT CONSIDERATIONS
Start small : Pine Script is not a GPU training environment. Compact models are the right starting point.
Control chart history : Use `calc_bars_count = 600` in `indicator()` when needed to balance available training history against model size and execution time.
Use the training gate : For heavier models, use `.withTrainingGate(true)` so backpropagation runs only at the confirmed historical edge.
Scale your inputs : Raw market features often differ by orders of magnitude. Use dataset scalers unless you have a deliberate reason not to.
Validate separately : Use `trainBatch()` and `validationBatch()` to monitor generalization instead of only watching training loss.
Avoid lookahead : Build feature rows only from information available at the time of the row. Use completed target rows for training.
Treat outputs as research signals : NeuraLib provides model mechanics. Strategy design, risk management, and market assumptions remain the user's responsibility.
----------------------------------------------------------------------------------------------------------------
🔷 API REFERENCE
🔸 Model Setup
sequential(name) : Creates an empty `Sequential` model.
compileConfig() : Creates a model configuration object.
build(rng) : Builds model parameters with a deterministic random stream.
compile(config) : Builds the model when needed and applies the training configuration.
rng(seed, streamId) : Creates a deterministic random stream.
🔸 Sequential Methods
input(dimsArray, name) : Defines the input shape.
dense(units, activation, name) : Adds a fully connected layer.
qHead(actionCount, activation, name) : Adds a Q-value output head.
activation(activationKind, alpha, name) : Adds an activation block.
dropout(rate, name) : Adds dropout regularization.
layerNorm(name) : Adds layer normalization.
flatten(name) and reshape(outputDimsArray, name) : Adjust model shape metadata.
block(graphBlock) : Adds a custom `GraphBlock`.
trainOnBatch(inputTensor, targetTensor) : Runs training when the active gate allows it.
backward(targetTensor) : Accumulates gradients from the last forward pass without stepping.
step() : Applies the optimizer step to accumulated gradients.
predict(inputTensor) : Runs inference.
predictMC(inputTensor, samples) : Runs dropout-enabled Monte Carlo prediction and returns mean and variance.
evaluate(inputTensor, targetTensor) : Calculates loss without updating weights.
fitDataset(dataset) and fitRollingDataset(dataset, targetOffset) : Train through dataset adapters.
getWeightsArray() and setWeightsArray(weightsArray) : Export and import flat model weights.
softUpdateFrom(sourceModel, tau) : Soft-update parameters from another model.
🔸 CompileConfig Methods
optimizer(optimizerState) : Sets the optimizer.
schedule(scheduleState) : Sets the learning-rate schedule.
loss(lossKind) : Sets the training loss.
reduction(reductionKind) : Sets loss reduction behavior.
metric(metricKind) : Adds a metric.
batchSize(size) , epochsPerBar(count) , evalStride(stride) , and historyLength(length) : Store batch and cadence preferences, and set the metric history length.
clipNorm(value) and clipValue(value) : Apply gradient clipping.
gradAccumSteps(steps) : Accumulates gradients before stepping.
withTrainingGate(enabled) : Restricts training to the last confirmed historical bar when enabled.
trainEveryCall() : Allows training whenever `.trainOnBatch()` is called.
presetPriceRegression() , presetReturnRegression() , presetBinaryDirection() , presetBinaryDirectionLogits() , presetQValues() , and presetSharpe() : Apply common loss and metric presets.
🔸 Datasets
windowDataset(featureCount, targetCount, maxRows, name) : Stores flat feature and target rows.
rollingDataset(timeSteps, featureCount, targetCount, maxRows, name) : Stores time-series windows.
pushRow(featureArray, targetArray) : Adds one validated row.
pushBuilderRow(featureBuilder, targetArray) : Adds a row from a `FeatureBuilder`.
pushNextReturnRow(featureBuilder, currentValue, futureValue) : Adds a next-return target.
pushNextDirectionRow(featureBuilder, currentValue, futureValue, threshold, zeroOne) : Adds a direction target.
ready(minRows or minWindows, targetOffset) and size() : Check dataset readiness.
lastBatch(batchSize) : Returns the most recent scaled rows from a `WindowDataset`.
toBatch() : Returns all rows from a `WindowDataset`.
unrollBatch(targetOffset) : Returns all rolling windows from a `RollingDataset`.
trainBatch(validationRows or validationWindows, targetOffset) : Returns the training side of the split.
validationBatch(validationRows or validationWindows, targetOffset) : Returns the validation side of the split.
setInputScaler(kind) , setTargetScaler(kind) , scaleInput(tensor) , scaleTarget(tensor) , and inverseScaleTarget(tensor) : Configure and apply scaling.
clear() : Clears stored rows.
🔸 Tensor, Matrix, and Feature Helpers
scalar(value) , vector(valuesArray) , matrix2d(rows, cols, fillValue) , zeros(shape) , ones(shape) , and full(shape, fillValue) : Create tensors.
shapeFromDims(dimsArray) : Creates a shape.
matrixTensor(tensor) , matrixTensor2d(rows, cols, fillValue) , and matrixTensorFromMatrix(sourceMatrix) : Create matrix tensors.
reshape(dimsArray) , flatten() , row(rowIndex) , get1d(index) , sum() , mean() , variance() , normL2() , argmax() , and dot(other) : Tensor methods.
matmul() , transpose() , add() , subtract() , multiply() , divide() , scale() , activate() , softmax() , sliceRows() , sliceCols() , concatRows() , concatCols() , globalAvgPool1d() , and globalMaxPool1d() : MatrixTensor methods.
featureBuilder(name) , push(value, featureName) , addFeature(value, featureName) , toTensor(tensorName) , toArray() , size() , and clear() : Feature row helpers.
🔸 Scalers, Optimizers, and Schedules
zScoreScaler() , minMaxScaler() , runningZScoreScaler() , and noneScaler() : Standalone scaler states.
fit(tensor) , partialFit(tensor) , transform(tensor) , and inverseTransform(tensor) : Scaler methods.
sgd(learningRate) , momentum(learningRate, momentum) , rmsprop(learningRate, rho, epsilon) , adam(learningRate, beta1, beta2, epsilon) , and adamW(learningRate, beta1, beta2, epsilon, weightDecay) : Optimizers.
constantSchedule(learningRate) , stepDecay(baseLearningRate, decaySteps, gamma) , cosineDecay(baseLearningRate, minLearningRate, decaySteps) , and warmupCosine(baseLearningRate, minLearningRate, warmupSteps, decaySteps) : Schedules.
currentRate(stepCount) : Reads a schedule's learning rate at a step.
paramBank() , append() , zeroGrad() , globalGradNorm() , step(optimizerState) , and softUpdateFrom(sourceBank, tau) : Low-level parameter bank utilities.
🔸 Losses and Metrics
mse() , mae() , huber() , logCosh() , binaryCrossEntropy() , binaryCrossEntropyFromLogits() , categoricalCrossEntropy() , softmaxCrossEntropyFromLogits() , directionalLoss() , quantileLoss() , multiHorizonWeighted() , and sharpeLoss() : Direct loss helpers.
metricValue(metricKind, predictionTensor, targetTensor) : Direct metric helper.
earlyStopper(patience, minDelta) , update(validationLoss) , and reset() : Validation stopping helper.
nextReturnValue(currentValue, futureValue) and nextDirectionValue(currentValue, futureValue, threshold, zeroOne) : Common target helpers.
🔸 GraphBlock Operations
graphBlock(name) : Creates a custom trainable graph block.
input() , param() , constScalar() , constMatrix() , and output() : Define graph inputs, parameters, constants, and output metadata.
matmul() , add() , subtract() , multiply() , divide() , scale() , activate() , softmax() , transpose() , layerNorm() , and dropout() : NeuraLib graph math.
causalMask() , sliceRows() , concatRows() , sliceCols() , concatCols() , reduceRows() , and reduceCols() : Structural graph operations.
globalAvgPool1d() , globalMaxPool1d() , attentionScore() , attentionApply() , conv1d() , scanLstm() , and scanGru() : Sequence and architecture primitives.
🔸 NeuraLib_Models API
prioritizedReplayBuffer(featureCount, targetCount, maxRows, name) : Creates a replay buffer.
pushExperience(featureRowArray, targetRowArray, priority) , sampleBatch(batchSize, alpha, beta, seed) , updatePriority(index, priority) , toBatch() , ready(minRows) , size() , and clear() : Prioritized Experience Replay helpers.
pushPositionalEncoding(position, dimensions, maxPeriod, featurePrefix) : Adds positional encoding values to a `FeatureBuilder`.
residualDense() , duelingQHead() , conv1d() , temporalConvStack() , globalAvgPool1d() , globalMaxPool1d() , lstm() , gru() , selfAttention() , multiHeadSelfAttention() , crossAttention() , transformerEncoder() , transformerEncoderStack() , and transformerDecoder() : NeuraLib_Models `Sequential` methods.
NeuraLib is for Pine Script developers who want to move beyond fixed formulas and experiment with real neural network workflows directly inside PulseWire. It is a research framework, not a guarantee of market performance. Use validation, avoid lookahead, control risk, and keep models small enough for Pine's execution limits.
All the diagrams in this publication are rendered natively on PulseWire using Pine3D
----------------------------------------------------------------------------------------------------------------
This work is licensed under (CC BY-NC-SA 4.0) , meaning usage is free for non-commercial purposes given that Alien_Algorithms is credited in the description for the underlying software. For commercial use licensing, contact Alien_Algorithms
Library
Liquidity Surge Forecast with Markov Chains [TechnicalZen]Clear direction from Markov Chains confirmed projections.
Publishing this v3 with all the enhancements users desired and more. Thank you for your feedback.
What This Is
A 3D liquidity-and-momentum visualization that tells you where the market is heading right now, how long the current state is likely to hold, and when the next regime change is expected — all backed by a 2nd-order Markov chain that learns from your chart's own history.
Two independent systems — Money Flow (MFI-driven) and Price Current (Hull-VWMA or signed-ADX) — render as layered dotted carpets inside a bounded 3D box. When both systems agree on direction AND the Markov chain confirms, a whale surfaces — 🐳 bullish, 🐋 bearish. Chop gets a shark 🦈. Sideways drift gets a crab 🦀. And when the Markov chain predicts an imminent regime transition, a small hatchling whale appears before confluence forms.
The Current State row tells you, in one line, exactly what to expect next.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Built On
Money Flow Dynamics Forecaster 3D — the original 3D layered-terrain architecture, MFI/RSI momentum carpet, Hull-VWMA price current carpet, slope-extrapolated forecast, rider + whale system.
Same 3D engine. Same dual-system confluence as the foundation. Then: regime classification, Markov statistical learning, current-state intelligence, and a live win-rate scoreboard on top.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Clear Direction — At A Glance
Most indicators show you lines and ask you to interpret. This one tells you plainly, in a single dashboard row:
What regime you're in right now — 🐳 Bull, 🐋 Bear, 🦈 Chop-zone, or 🦀 Sideways
How long it's been held — in bars
Whether the regime is BALANCED or IMBALANCED — based on the Markov chain's next-bar probabilities
When the next regime change is expected — in bars, computed from the stay-probability
Which direction the market leans next — the highest-probability non-current state
Example readouts:
"Current State: 🐳 Bull held 5b · IMBALANCED — stay 68%, change expected in ~3b · next lean: 🦀 Sideways"
"Current State: 🦀 Sideways held 12b · BALANCED — change imminent · next lean: 🐳 Bull"
"Current State: 🐋 Bear held 2b · IMBALANCED — stay 82%, change expected in ~5b · next lean: 🦀 Sideways"
No interpretation required. You read the line, you know where you are, you know what to expect.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
New Features — And Why Each Exists
1. Current State intelligence row
Why: Confluence indicators tell you WHEN a signal fires. They don't tell you "how solid is the current regime," "is a change coming," or "how long do I have before conditions flip." The Current State row answers all three in one glance.
The balanced / imbalanced distinction matters most:
BALANCED — the three next-bar probabilities are close to 1/3 each. No clear direction. A regime change is imminent (could go anywhere). Trade lighter, wait for resolution.
IMBALANCED — one direction dominates. Regime has a preferred path. The stay-probability tells you how long it's likely to persist; the next-lean tells you which direction it will tilt when it does flip.
The expected-bars-to-change is a geometric distribution mean: 1 / (1 − P(stay)). If a regime has 80% stay probability, it's expected to persist ~5 more bars. If 33%, it's expected to flip in ~1.5 bars.
2. 2nd-order Markov chain regime predictor
Why: the original whale logic was reactive — it fires after confluence forms. Markov is predictive — it learns your instrument's transition habits and uses them to gate and anticipate whale signals.
Pure statistics, no black box:
2nd-order — predicts the next regime from the pair of previous regimes, not just one. Captures patterns like "Chop → Sideways → 68% Bull next" that a 1st-order chain would miss.
Laplace smoothing (α=1) — every transition count gets a +1 pseudocount before normalization. Prevents "never observed → 0% forever" failure. Standard in real statistics.
Exponential recency decay — newer transitions count more than old ones (default 0.995/bar). Markets drift; stale history shouldn't dominate current prediction.
Duration conditioning — separate transition matrices for "current state held <5 bars" vs "held ≥5 bars." Regimes behave differently after they've been running. Real statistical sub-populations.
Confidence gating — if the current context has fewer than 10 observations, predictions are flagged low-n . No fabricated probabilities.
Maximum useful substance without gimmick. 3rd-order Markov would need thousands of regime transitions per cell to converge — doesn't happen on typical charts. 2nd-order is the ceiling before diminishing returns.
3. Dual-layer regime classification — Chop-zone 🦈 vs Sideways 🦀
Why: prior versions treated "not trending" as a single category. But there are two fundamentally different kinds of non-trending market:
🦈 Chop-zone — violent range-bound circling. Detected via classic Choppiness Index . Often precedes a sharp breakout.
🦀 Sideways — slow drift, flat angles across close/high/low at both short and long periods. Detected via angle consensus . Often indicates accumulation or distribution.
Showing them separately lets you read which kind of non-trending you're in. Different implications, different decisions.
4. Hatchling whales — pre-signal pre-whales
Why: the Markov chain lets us anticipate confluence before it forms. When the current state is Sideways AND Markov predicts Bull (or Bear) with confidence above the hatchling threshold, a small whale appears at the mid-forecast position — a heads-up that confluence is probabilistically coming.
Full whale (size.huge at forecast edge) = confluence is here now.
Hatchling whale (size.small at forecast mid) = confluence is probably coming soon.
Better entries on regime changes.
5. Markov-gated whale confirmation
Why: sometimes projected-confluence fires, but the instrument's historical pattern says "from this context, the opposite is more likely." That's exactly the setup a trader wants the indicator to filter out .
The gate is permissive — Markov blocks a whale only if it's confident AND its argmax points the opposite direction. Uncertainty or agreement = pass through. Reduces false confluence without over-filtering.
6. Regime-aware 4-column win-rate dashboard
Why: knowing how much time the instrument actually spends in each regime is as actionable as the signals themselves.
Four parallel columns:
🐳 Bull — confluence signals and win rate
🐋 Bear — same, opposite direction
🦈 Chop-zone — CI chop events and % of bars
🦀 Sideways — angle-sideways events and % of bars
If your instrument spends 80% of bars in Chop/Sideways, confluence will be rare — adjust timeframe or instrument. If Markov shows low-n on most bars, the matrix isn't populated yet — wait for more history before trusting predictions.
7. Session-aware for futures
Why: NQ, ES, CL and other overnight-session futures stamp their daily bar at session start , which is the previous calendar evening. Naive `dayofweek(time)` reads NQ's "Friday session" as Thursday. The indicator uses `time_close("D")` — the close of the daily bar, always on the trading date — so regime classification is correct for both cash equities (TSLA, SPY) and overnight futures (NQ, ES). Same indicator, any asset class.
8. Honest evaluation — next-signal MFE or directional close
Why: "close-at-N-bars" is dishonest. Price can move 2×ATR favorably then retrace — close-at-N logs that as a loss. MFE logs it as what it actually was.
Each signal is held pending until the next signal fires. It's a win if either:
The close at next-signal bar was directionally favorable vs entry, OR
The Maximum Favorable Excursion between the two signals reached the ATR-scaled threshold (default 0.5×ATR at entry bar)
Either qualifies. Transparent. Computed live. Disclaimer embedded in the dashboard footer.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
How to Read the Dashboard
┌────────────────────────────────────────────────┐
│ Liquidity Surge + Markov · Win Rate │
├────────┬────────┬─────────────┬───────────────┤
│🐳 Bull │🐋 Bear │🦈 Chop-zone │🦀 Sideways │
│42 sigs │38 sigs │7 events │12 events │
│31 wins │24 wins │120 bars │45 bars │
│73.8% │63.2% │23% │8.6% │
├────────────────────────────────────────────────┤
│Markov Forecast Next: 🐳 52% · 🦀 31% · 🐋 17% │
├────────────────────────────────────────────────┤
│Current State: 🐳 Bull held 5b · IMBALANCED │
│ stay 68%, change expected in ~3b · next: 🦀 │
├────────────────────────────────────────────────┤
│⚠ Not financial advice · Learned on chart hist │
└────────────────────────────────────────────────┘
Reading order:
Column data — how Bull/Bear signals have performed, how much time is spent in each regime
Markov Forecast Next — next-bar regime probabilities (with sample-size confidence)
Current State — the single-line answer to "where am I and what's next"
Footer — disclaimer + config
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
How to Use
Load with defaults.
Wait for ~100-200 bars of history. The Markov matrix needs observations to learn.
Read Current State first. It tells you what to expect.
If BALANCED → expect a regime change, trade light.
If IMBALANCED + change in ~N bars → plan around that window.
Watch for 🐳 / 🐋 full whales at the forecast edge — confluence + Markov confirmed.
Watch for small hatchling whales at forecast mid-point — Markov's early prediction of confluence coming.
Respect 🦈 (chop-zone) and 🦀 (sideways). Don't fight the regime.
Tune Hatchling Threshold and Win Threshold (×ATR) to your instrument and style.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Best Paired With Smart Candle Structures
This indicator answers whether and when to trust the flow. Smart Candle Structures answers where to act. Together: right place, right moment, measurable conviction, regime-aware.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Key Settings
Time Span — past bars rendered + forecast horizon (default 15)
Momentum Source — MFI (default) or RSI
Oscillator Type — Hull-VWMA (default) or signed-ADX
Slope Lookback — bars for slope that fires whales (default 4)
Evaluation Window — bars after a signal to measure MFE (default 5)
Win Threshold (× ATR) — minimum favorable excursion as a multiple of ATR (default 0.5)
Gate Whale by Choppiness — master toggle for 🦈 / 🦀 filter
CI Length / CI Threshold — classic Choppiness Index tuning
Angle Short / Long Period — angle-based sideways lookbacks
Angle Trend / Sideways Threshold — angle degrees defining trending vs sideways
Use Markov Predictor — master toggle for the 2nd-order chain
Count Decay per Bar — recency weighting for Markov counts (default 0.995)
Hatchling Threshold — minimum Markov probability to fire pre-whale (default 55%)
Dashboard Text Size — Tiny / Small / Normal / Large / Huge
Camera — yaw / pitch / scales for the 3D view
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Disclaimer
This is a visualization and analytical tool, not financial advice or a signal service. The Markov chain is trained on your chart's history — it describes what has happened on this instrument at this timeframe, not what will happen. Regime transition probabilities are learned estimates; past frequencies do not guarantee future outcomes. Markets are reflexive and can transition in ways the chain has never observed. Hatchlings, whales, sharks and crabs are visualizations of mathematical predictions — they do not constitute buy or sell recommendations. Trade with your own risk management. Every trade can lose.
The indicator echoes this disclaimer in its dashboard footer so you see it every time you read the chart. It's always there because it's always true.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Clear direction from learned regimes.
— TechnicalZen
Indicator
Uptrick: ML Kernel Regression
Introduction
This indicator applies Nadaraya-Watson kernel regression, a non-parametric machine learning estimator, directly to price data in order to produce a smooth, noise-reduced representation of the market's underlying trend. Unlike moving averages that apply equal or linearly decaying weights, this method uses a Gaussian kernel function to assign weights based on how far back in time each bar sits relative to the current one. Bars closer in time receive exponentially higher weights, while older bars decay naturally. The result is a regression curve that adapts organically to local price structure rather than imposing a fixed lag model onto the data. Residual bands are then constructed around this curve using the rolling standard deviation of the difference between price and the regression line, forming dynamic envelopes that reflect actual price dispersion rather than arbitrary multipliers of a fixed moving average.
The indicator is designed for traders who want a statistically grounded trend baseline with state-driven directional signals, without relying on lagging traditional averages. It is built in Pine Script v6 and is fully non-repainting. All state decisions are committed only on confirmed bars, meaning no signal is generated intra-bar and no future bar data influences the output.
How It Works
The core calculation is the Nadaraya-Watson estimator. At each bar the indicator looks back across a user-defined window and computes a weighted average of past closing prices. The weight assigned to each historical bar is determined by the Gaussian kernel: weight = exp( -lag² / (2 · h²) ), where lag is the number of bars back and h is the bandwidth parameter. A larger bandwidth makes the curve smoother and slower to react. A smaller bandwidth makes it more reactive but noisier.
When adaptive bandwidth is enabled, the bandwidth h is scaled dynamically by a normalised ATR factor. In volatile periods, the kernel widens, producing a smoother estimate that avoids overreacting to spike conditions. In calm periods, the kernel tightens, allowing the curve to track price more closely. This makes the regression inherently context-aware without requiring the user to manually switch settings across different market regimes.
The residual at each bar is defined as the difference between the closing price and the regression value. A rolling standard deviation of these residuals forms the sigma value, which is then smoothed via EMA. The upper and lower bands are placed at a user-controlled multiple of sigma above and below the kernel line. Because the bands are derived from actual price-to-regression deviation, they expand during high-dispersion conditions and contract when price tracks the regression tightly.
State is classified as bullish when price closes above the upper band on a confirmed bar, and bearish when price closes below the lower band on a confirmed bar. Between breakout events the state persists, meaning the indicator holds its last valid directional reading rather than flipping to neutral. This gives the signal a regime-like quality rather than a purely oscillatory one.
Features
Nadaraya-Watson Gaussian kernel regression curve computed from scratch over a fully user-controlled lookback window
Adaptive bandwidth scaling driven by a normalised ATR factor, widening the kernel during volatile conditions and tightening it during calm ones
Residual-based standard deviation bands that expand and contract with actual price-to-regression dispersion rather than fixed multipliers
Smoothing controls for both the main regression output and the band width, allowing fine-tuning of reactivity versus stability
Three visual display modes: Bands mode showing the full envelope, Single Line mode showing only the regression curve with a gradient fill toward price, and Trail mode showing only the relevant band side as a directional trail
Gradient fills in all three visual modes that fade from the regression line outward toward price, maintaining visual clarity without obscuring price action
State-based bar coloring that applies the directional regime color to every candle, using custom plotcandle rendering for full wickcolor and bordercolor consistency
Signal labels that appear only on confirmed state transitions, placed at user-selectable anchors including High or Low, the Main Line, or the Band levels, with adjustable ATR-based offset
Seven selectable color themes covering Classic, Cyber Aqua, Crimson Pulse, Royal Purple, Emerald Night, Minimal Mono, and Classic Emerald, each providing a complete set of bull, bear, neutral, background, and frame colors
A live dashboard table displaying current signal direction, kernel MA value, upper band value, lower band value, current sigma width, and active bandwidth including whether adaptive mode is engaged
Alert conditions for bullish breakout above the upper band and bearish breakdown below the lower band, both tied to confirmed crossover and crossunder events
Toggle controls for bar coloring, band fill, and the dashboard table independently
Dashboard:
Band Mode:
Single Line Mode:
Trail Mode:
Inputs
Lookback Window: controls how many historical bars the Gaussian kernel sums over. Larger values produce a slower, broader regression curve. Default is 30.
Base Bandwidth (h): sets the core width of the Gaussian kernel. Higher values create smoother, more generalized curves. Lower values track price more closely. Default is 8.0.
Adaptive Bandwidth: when enabled, the bandwidth is multiplied by a factor derived from normalised ATR, making the kernel wider in volatile conditions. Default is enabled.
ATR Length (adaptive): the period used to compute the ATR for adaptive scaling. Default is 14.
MA Output Smoothing: applies an EMA pass over the raw regression output to reduce micro-jitter in the curve. Default is 3.
Band Multiplier (sigma): how many standard deviations above and below the regression line the bands are placed. Default is 1.0.
Band Lookback (sigma): the rolling window used to compute the standard deviation of residuals. Default is 24.
Band Smoothing: EMA smoothing applied to the raw sigma value to stabilize band movement. Default is 5.
Visual Mode: selects between Bands, Single Line, and Trail display modes.
Color Bars: enables state-colored candles. Default is enabled.
Fill Bands: enables the semi-transparent fill between upper and lower bands. Default is enabled.
Show Dashboard: toggles the live data table. Default is enabled.
Color Gradient Smooth: controls color smoothing, currently reserved for future gradient transitions.
Label Anchor: selects where signal labels are pinned. Options are High or Low, Main Line, and Bands.
Offset Mult (ATR): scales how far above or below the anchor point labels are offset. Default is 0.50.
Theme: selects the color theme across all visual elements.
Alert: Cross Above Upper Band: enables the bullish breakout alert condition.
Alert: Cross Below Lower Band: enables the bearish breakdown alert condition.
Originality
The originality of this script lies in the combination of a properly implemented Nadaraya-Watson estimator with an ATR-adaptive bandwidth system, residual standard deviation bands, and a persistent non-neutral state engine, all packaged with a multi-mode visual system that adjusts its presentation to the current directional regime. The regression curve is not a modified moving average. It is a genuine weighted least squares estimate computed bar by bar using a Gaussian kernel function. The adaptive bandwidth mechanism means the indicator does not treat all market conditions equally, which is a meaningful departure from static-parameter band systems. The state logic prioritises confirmed readings and persists between band contacts, which makes the regime classification stable and avoids the false-neutral problem common in threshold-based indicators. The three visual modes serve distinct use cases: Bands for envelope and breakout context, Single Line for a clean trend baseline, and Trail for a dynamic support or resistance reference that follows the active regime. These elements are not assembled from existing published open-source scripts; the full codebase is original work by the author.
Conclusion
Uptrick: ML Kernel Regression provides a statistically grounded approach to price smoothing and trend regime classification by applying a Gaussian kernel estimator rather than a conventional moving average. The adaptive bandwidth, residual bands, and persistent state logic work together to give traders a tool that reflects actual market behaviour rather than imposing fixed parameters onto it. The multiple visual modes and theme system make it practical across a range of chart styles and use cases.
Disclaimer
This script is published for educational and analytical purposes only. Nothing in this script or its description constitutes financial advice, investment advice, or a recommendation to buy or sell any asset. All trading involves risk. Past performance of any indicator or signal does not guarantee future results. You are solely responsible for your own trading decisions.
Indicator
ML Trend Architect [TechnicalZen]ML Trend Architect
A regime visualization and forecasting engine that combines machine learning confidence scoring with curved 3D glass facades to map market structure and anticipate regime transitions in real-time.
What It Does
This indicator identifies directional regime changes using a dual-engine approach — an impulse momentum engine and a DX/ADX directional movement engine — then wraps them in curved 3D visual envelopes that show regime strength, direction, and conviction at a glance. Beyond detection, it actively forecasts regime quality by scoring each transition against historical analogs and projecting follow-through probability before the move fully develops.
How It Works
Regime Detection
Band-break events (flips, continuations, refreshes) drive regime transitions. Price breaking above the upper band triggers a bullish flip; below the lower band triggers bearish. The system requires confirmation bars and enforces cooldowns to prevent whipsaws.
Forecasting Engine
At each regime transition, the engine evaluates how likely the new regime is to follow through before the move has played out. It scores setup quality across multiple dimensions — anchor cluster positioning, momentum alignment, compression state, duration, excursion depth, and family coherence — to produce a forward-looking conviction percentage. When KNN Memory is active, it deepens this forecast by matching the current setup fingerprint against resolved historical analogs, asking: "of all similar setups in the past, what percentage actually followed through?" The result is a probabilistic forecast — not a reactive signal — displayed as a confidence label at the moment of regime change.
Forward Resolution and Accountability
Every forecast is tracked forward over a configurable evaluation window. The system monitors whether price held on the correct side of the anchor cluster and whether it expanded sufficiently. Forecasts that fail to materialize are marked with a rejection symbol — creating a visible track record of forecast accuracy directly on the chart. This closed-loop accountability distinguishes it from indicators that fire and forget.
Curved Glass Facades
Instead of flat rectangular boxes, regime segments are rendered as curved 3D envelopes using square-root decay from the band edges. The curves start wide at segment birth and taper naturally, creating organic shapes that reflect each regime's character — steep narrow curves for aggressive moves, wide gradual sweeps for grinding trends.
DX Heat Strip
A color-gradient ribbon flows along the curved walls showing directional conviction in real-time. Yellow indicates weak/choppy conditions; cyan indicates strong bullish momentum; magenta indicates strong bearish pressure. Bulls support from below, bears press from above.
4-Color Regime System
The regime uses a continuous gradient driven by ADX position and slope, producing four distinct states that blend smoothly:
- Teal: strong bullish momentum (ADX positive, rising)
- Orange: fading bullish momentum (ADX positive, falling)
- Light Green: fading bearish pressure (ADX negative, rising)
- Red: strong bearish pressure (ADX negative, falling)
Signal Leader Detection
The engine tracks which signal source — Impulse or DX — has been more accurate recently and who fires first. This is itself a forecast: it predicts which signal source you should weight more heavily in the current market regime. The dashboard displays the current leader with win rate and directional bias.
Connection Lines
Dashed lines connect same-direction hull phases across interruptions, drawing on highs (bear color) and lows (bull color). When the price connection slope diverges from the indicator slope, lines become solid and thicken — flagging potential divergence setups that often precede regime reversals.
The Forecasting Pipeline
Step 1: Context Assessment — Where is price relative to the 7-anchor cluster? Are anchors aligned (coherent) or scattered? Is the market compressed or extended?
Step 2: Trigger Quality — How decisive is the breakout bar? Body ratio, close location, force spread, and cluster clearance are scored.
Step 3: Regime Environment — ADX strength, relative volume, and ATR regime provide the macro backdrop.
Step 4: Analog Matching (KNN) — The current feature fingerprint is compared to historical setups. Nearest neighbors are found using normalized Manhattan distance across 5 feature dimensions. Their resolved outcomes (success/failure) are distance-weighted to produce a probability.
Step 5: Blended Forecast — Base score and KNN probability are blended (configurable weight) into a final conviction percentage.
Step 6: Forward Tracking — The forecast is monitored over the evaluation window. Hold ratio and expansion targets determine success or failure.
Step 7: Memory Update — Resolved outcomes feed back into the KNN memory bank, continuously improving forecast accuracy on the current instrument.
Theme Presets
Curved — Full curved glass facades with 3D depth, regime fills, and heat strip
Classic — Clean curved envelopes without 3D effects
Heat Strip — Minimal view showing only the DX heat ribbon, markers, and connection lines
Custom — Full manual control over all visual parameters
Key Settings
My Trade Style — Balanced (standard width), Conservative (wider bands), or Aggressive (tighter bands)
Regime Mode — Swing (4-color ADX gradient) or Scalp (2-state impulse-driven)
Mid Line View — Lead (mid sits opposite price as support/resistance) or Follow (mid tracks toward price)
KNN Memory — Enable adaptive learning from historical pattern outcomes
Confidence Threshold — Minimum confidence % required to display a forecast label
Theme Preset — Quick visual style switching
Dashboard
The dashboard displays: Signal Leader (with win rate and direction), ADX Trend, Regime state, Structure (zone from VWMA positioning), Conviction (dominant confidence with ML/base tag), Impulse direction, Volatility regime, Session status, Win Rates, and Mode.
How to Use
Watch the regime color for the macro direction — teal and orange are bullish states, green and red are bearish
Read the heat strip color for conviction strength — vivid colors mean strong directional movement, yellow means caution
Follow the connection lines for structural trend tracking — solid thick lines with diamonds flag divergence between price and momentum
Trust confidence labels above your threshold — higher % means the forecast matches historically successful patterns
Respect rejection markers (x) — they mark where a forecast failed to materialize
Check the Leader in the dashboard — trade with the signal source that has been more reliable recently
Enable KNN Memory for instruments you trade regularly — it improves forecast accuracy as it accumulates resolved outcomes
Technical Notes
Works on any instrument and timeframe
No repainting — all signals confirm on bar close
KNN memory is session-scoped and builds over time on the current instrument
3D rendering uses polylines (max 100) with configurable segment count
Confidence scoring uses a 7-anchor cluster consensus, not a single indicator
Forecasts are forward-resolved with visible accountability — every prediction is tracked and scored
Disclaimer
This indicator is provided for educational and informational purposes only. It is not financial advice and should not be used as the sole basis for any trading or investment decision. Past performance of any signal, pattern, or scoring system does not guarantee future results. All trading involves risk, including the potential loss of principal. The machine learning components learn from historical data on the current instrument and timeframe — their predictions reflect pattern similarity, not certainty. The term "forecast" refers to probabilistic scoring of setup quality based on historical analogs, not a guarantee of future price direction. Always conduct your own analysis, manage your risk appropriately, and consult a qualified financial advisor before making trading decisions. The developer assumes no liability for any losses incurred through the use of this tool.
Indicator
AI SuperTrend Strength Forecasting Engine [TraderZen]The question that matters most in real time: "How confident should I be in this setup right now?"
The ML Trend Strength Forecasting Engine was built to answer that question directly. Instead of drawing lines on a chart and leaving interpretation to the trader, it computes a numerical confidence score for every potential bull and bear setup as it forms. It evaluates the quality of the setup context, the strength of the trigger bar, the state of the broader market regime, and whether similar setups in the recent past actually worked. The output is a percentage — how likely is this signal to follow through.
The indicator does not generate buy or sell signals in the traditional sense. It provides a calibrated confidence reading that helps the trader decide whether to act, wait, or pass entirely.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Key Differences and Advantages
What KNN Predicts
Generally: "Will the trend direction be right?"
This Indicator: "Will this setup hold or expand?"
Training Labels
Generally: Past indicator directions (circular — predicting itself)
This Indicator: Actual price outcomes (self-validating — did the trade work?)
Signal Source
Generally: Single trend line or oscillator
This Indicator: 7-anchor cluster consensus (VWAP, VWMA, EMA, SMA, RMA, HMA, Donchian)
Features
Generally: RSI + MA deviation at multiple timeframes
This Indicator: Context + Trigger + Volume + ADX + Trend Distance — Kalman-filtered
Confidence Meaning
Generally: "How similar to past indicator wins"
This Indicator: "How likely to follow through based on setup quality + historical analogs"
Accountability
Generally: No rejection tracking, no performance feedback
This Indicator: Rejection markers on failed signals + live win rate dashboard
Adaptivity
Generally: Fixed normalization window, static thresholds
This Indicator: Adaptive ceilings, vol regime scaling, running stats, memory decay
How It Works
The engine runs five interconnected layers , each feeding into a final confidence score.
Layer 1 — The Anchor Cluster
Seven moving averages are computed simultaneously: VWAP, VWMA, EMA, SMA, RMA, HMA, and the Donchian midpoint. These are grouped into four families:
Institutional — VWAP and VWMA
Trend — EMA, SMA, RMA
Fast — HMA
Structural — Donchian midpoint
The high, low, and midpoint of this entire cluster define where the market's center of gravity sits. A bullish candidate is detected when price crosses above the cluster midpoint from below. A bearish candidate is detected when price crosses below from above. The crossing is the trigger event. Everything else evaluates the quality of that cross.
Layer 2 — Context Scoring
Before the cross happens, the engine has been watching:
How long price spent on the other side ( duration )
How far it traveled away from the cluster ( excursion )
How tightly the anchors are compressed ( compression )
How much slope agreement exists across anchor families ( coherence )
Where price sits relative to the cluster edges ( clearance )
These components combine into a context score that answers: "Was the setup that led to this cross a good one?"
Context scoring is weighted — side positioning and slope carry the most influence, followed by compression and clearance, with duration and excursion providing secondary confirmation.
Layer 3 — Trigger Scoring
The crossing bar itself is evaluated for quality. A strong trigger bar:
Has a solid body (not a doji)
Closes near the directional extreme (close near the high for bulls, near the low for bears)
Shows force momentum aligned in the right direction
Has meaningful distance from the cluster
Represents a genuine cross rather than a wick touch
Each dimension is scored and blended into a trigger score.
Layer 4 — Regime and Features
The broader environment is assessed through Kalman-filtered features : relative volume versus its baseline, relative ATR versus its baseline, ADX trend strength, and price distance from the trend moving average. These are smoothed to reduce noise while preserving directional shifts.
Volatility regime awareness scales key thresholds automatically. In low-volatility environments, the engine tightens its excursion and distance requirements. In high-volatility environments, it loosens them. This prevents the indicator from being too sensitive in calm markets and too restrictive in active ones.
Layer 5 — Adaptive Analog Memory (KNN)
This is the differentiator. Every time a candidate signal is generated, the engine stores its five feature values (context, trigger, volume, ADX, trend distance) along with what actually happened over the following evaluation window. Did price hold above the cluster? Did it expand meaningfully? The outcome is recorded as a success or failure.
When a new candidate appears, the engine searches its memory for the most similar historical setups using a k-nearest-neighbors algorithm. Key details:
Features are normalized to equal scales using running statistics
Older samples receive exponentially decaying weights so the memory adapts as market behavior shifts
The KNN prediction is blended with the base confidence score
The memory learns and improves as it accumulates data on the specific instrument and timeframe
The memory requires a warmup period. Until enough resolved candidates have been recorded, the engine relies solely on the base scoring.
Putting It Together
The final confidence score blends the base score (context, trigger, regime, trend alignment) with the KNN analog prediction. The result is displayed as a percentage for both the bull and bear side. When confidence exceeds the threshold and beats the opposite side by a directional edge, a signal label appears on the price chart.
If a signal later fails to hold or expand, a small rejection marker ("x") appears on the chart — providing direct visual accountability.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
How To Use It
Reading the Pane
The indicator occupies its own pane below the price chart. Five elements are displayed:
Ribbon (top bar) — shows current state by color:
Gray = idle, Yellow = watching bull, Purple = watching bear, Aqua = armed bull, Fuchsia = armed bear, Green = bull signal, Red = bear signal
Dominant confidence line (thicker) — tracks the higher of the two side scores. When it rises above the confidence threshold line, a signal is likely imminent.
Bull and bear confidence lines (thinner) — show each side independently. Watching these diverge helps anticipate which direction will win.
Bias line — oscillates around 50. Above 50 = bull dominant. Below 50 = bear dominant.
Reading the Chart Labels
Green label with % — bull signal fired with confidence above threshold
Red label with % — bear signal fired
Dimmed green "x" — a bull signal was rejected (failed to hold or expand)
Dimmed red "x" — a bear signal was rejected
Higher percentages indicate stronger setups. Every signal is eventually graded pass or fail.
Reading the Dashboard
The dashboard in the bottom-right corner shows:
Bull and bear signal counts and win rates
Memory status (warming up or active with sample counts)
Current volatility regime (low, normal, or high)
Session status and bar count
Context and trigger score breakdown
Family coherence percentage
Force momentum direction
The win rate is the most important dashboard metric. Above 60% = well-calibrated. Below 45% = thresholds may need adjustment.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Important Settings
Confidence Threshold (default: 65%)
The minimum confidence required for a signal label to appear. This is the primary sensitivity control . Lower = more signals with weaker setups. Higher = fewer, more selective signals.
Armed Threshold (default: 45%)
Minimum confidence for the ribbon to show an armed state (aqua/fuchsia). Acts as a pre-signal alert — when the ribbon transitions from watch to armed, a signal may be imminent.
Watch Threshold (default: 45%)
Minimum context score to enter watch mode (yellow/purple). Lower values start tracking setups earlier.
Directional Edge (default: 5.0)
Minimum gap between bull and bear confidence required for a signal. Prevents signals in ambiguous conditions where both sides score similarly.
Resolution Bars (default: 8)
How many bars after a signal the engine waits before grading it pass or fail. Shorter = more responsive grading. Longer = more forgiving.
Hold Outcome Ratio (default: 0.60)
Fraction of resolution bars price must stay on the correct side for a success grade. At 0.60 with 8 bars, price must hold for at least 5 of 8 bars.
Expansion Outcome Target (default: 1.00 ATR)
Minimum price expansion required for an alternative success grade. A signal can succeed by either holding position or expanding sufficiently .
Memory Decay Rate (default: 0.005)
How quickly older memory samples lose influence. Higher = more recency-biased. At default, recent samples carry roughly 2x the weight of samples from 140 bars ago.
History Size (default: 120)
Maximum resolved samples in analog memory. Larger = more context but more computation.
Session Settings
Configure to match the instrument's primary trading session. Default is 0930-1600 America/New_York (US equities and futures).
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
What This Indicator Is Not
It does not predict price direction. It evaluates setup quality and follow-through probability.
It does not replace risk management . A 75% confidence signal can and will fail. The rejection markers exist to make this visible.
It does not work equally on all instruments without adjustment. The adaptive mechanisms handle much of the calibration automatically, but the confidence threshold may need tuning. The dashboard win rate provides the feedback loop.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Disclaimer
This indicator is provided for educational and informational purposes only. It is not financial advice, and it does not constitute a recommendation to buy, sell, or hold any financial instrument.
All trading involves risk. Past performance of any signal, scoring system, or pattern recognition mechanism does not guarantee future results. The confidence percentages represent a statistical assessment based on the scoring model and historical analogs available within the loaded chart data. They are not predictions and should not be treated as certainties.
The adaptive analog memory learns from the chart data currently loaded in PulseWire. Its effectiveness depends on having sufficient resolved samples, and its learned patterns may not generalize to future market conditions, different instruments, or different timeframes.
The win rate displayed in the dashboard reflects performance on the loaded chart history only and is subject to survivorship bias, lookback bias, and data limitations inherent to backtesting on historical bars.
No indicator, algorithm, or model can account for all market variables including liquidity events, news-driven gaps, exchange outages, or sudden regime changes. Traders should always use independent risk management, position sizing, and their own judgment before entering any trade.
By using this indicator, you acknowledge that you are solely responsible for your own trading decisions and that the authors accept no liability for any losses incurred.
Indicator
Indicator
Hidden Markov Model: Baum-Welch [UAlgo]Hidden Markov Model: Baum-Welch is a regime detection and reversal signaling indicator that applies a 3 state Hidden Markov Model to normalized log returns and continuously adapts its parameters using an online Baum Welch expectation maximization routine. The script is designed to classify the market into three latent regimes, then express that classification as real time probabilities for Bull, Range, and Bear conditions.
The indicator runs in its own pane ( overlay=false ) and outputs:
Probability curves for the three regimes
A dominant regime score scaled to 0 to 1
A regime strip visualization for quick bias reading
Adaptive background coloring based on the dominant regime and confidence
Optional regime shift markers
Optional buy and sell reversal markers driven by strict multi condition logic
The core idea is that price behavior can be modeled as transitions between hidden states that each have their own return distribution. The script fits a Gaussian emission model for each state, estimates state transition probabilities, and updates the posterior probability of each state on every bar. It retrains the full model at fixed intervals, while using a faster one step forward update between retrains for efficiency.
This implementation is not a simple threshold oscillator. It is a full mini HMM engine built in Pine with:
Scaled forward and backward algorithms
Expectation step producing gamma and xi posteriors
Maximization step updating initial distribution, transition matrix, state means, and state variances
Safeguards such as variance floors and transition floors to maintain numerical stability
The output is a regime aware probability system that can be used for bias, context, and reversal confirmation rather than simple entry signals.
Educational tool only. Not financial advice.
🔹 Features
🔸 1) Three State Hidden Markov Model Regime Engine
The model uses three hidden states and continuously estimates the probability of being in each state:
Bull regime
Range regime
Bear regime
This gives a probabilistic regime map rather than a single hard classification.
🔸 2) Baum Welch Training with Scheduled Retraining
The script retrains its parameters using an EM routine at a user defined interval in bars. Each retrain runs a configurable number of EM iterations. Between retrains, the indicator performs a one step forward Bayesian update of the posterior state probabilities.
This structure balances adaptability with performance.
🔸 3) Normalized Log Return Observations
The observation series is a z score normalized log return:
Log returns convert price changes into additive units
An EMA and rolling standard deviation normalize the series to stabilize the HMM fit
This helps the HMM learn regimes based on relative return behavior rather than raw price scale.
🔸 4) Automatic Bull, Range, and Bear Role Assignment
The model learns state means. The script then assigns roles by ranking those learned means:
The state with the lowest mean becomes the Bear state
The state with the highest mean becomes the Bull state
The remaining state is treated as Range
This keeps regime labeling consistent even as the internal state ordering shifts during training.
🔸 5) Probabilities and Dominant Regime Visualization
The script plots:
Bull probability curve
Range probability curve
Bear probability curve
It also plots an area for the dominant probability and a regime strip that makes it easy to see the dominant regime quickly without reading the full curves.
🔸 6) Regime Score Line (Bull minus Bear)
A continuous score is calculated as Bull probability minus Bear probability, then scaled to a 0 to 1 range. This score becomes the main regime momentum signal used for rebound and reversal logic.
🔸 7) Adaptive Background Coloring by Regime and Confidence
The pane background color changes based on the dominant regime. Transparency adapts according to confidence, so strong regime certainty produces a more visible background while low certainty remains subtle.
🔸 8) Strict Signal Filters for Bias and Reversal
The indicator provides bias filters:
Bull bias when Bull probability and confidence exceed thresholds and the dominant regime is Bull
Bear bias when Bear probability and confidence exceed thresholds and the dominant regime is Bear
It also provides reversal style buy and sell signals based on a multi condition framework described in the calculations section.
🔸 9) Reversal Logic Combining Extremes, Rebounds, and Transition Edge
Reversal signals are not generated by a single crossover. The script requires:
An extreme score pivot
An extreme regime probability at that pivot
A rebound trigger through predefined rebound levels
A minimum probability and confidence filter
A transition asymmetry and edge condition that favors switching toward the target regime
A momentum condition requiring Bull probability rising and Bear probability falling for buys, and the inverse for sells
A time window limit so reversals must occur within a limited number of bars after the extreme
This creates a high selectivity reversal engine.
🔸 10) Transition Matrix Insight and Switch Edge Metrics
The script computes predicted transition probabilities toward Bull and Bear using the current posterior and the transition matrix. It also measures transition asymmetry between Bull to Bear and Bear to Bull and uses these values as part of reversal confirmation.
This adds structural information that classic oscillators do not capture.
🔸 11) Anti Duplicate Reversal Signals
Once a pivot extreme has been used to generate a reversal signal, it is marked as consumed so the same pivot cannot repeatedly trigger additional buy or sell signals. This helps avoid signal repetition.
🔸 12) Full Informational Label Output
A live info label prints:
Current regime
Current signal text
Confidence
Bull, Range, Bear probabilities
Log likelihood
Key trigger thresholds
Reversal settings and edge settings
This provides transparency into what the model is currently seeing and why signals are or are not appearing.
🔹 Calculations
1) Observation Series: Normalized Log Returns
The script uses log returns:
logRet = math.log(close / nz(close , close))
Then normalizes them with an EMA mean and rolling standard deviation:
retMean = nz(ta.ema(logRet, normLength), 0.0)
retStd = math.max(nz(ta.stdev(logRet, normLength), 0.0), 1e-6)
obs = (logRet - retMean) / retStd
This creates an observation series with more stable scale properties across time.
2) Rolling Observation Window
The HMM is trained on a rolling window of length windowLen . Only the most recent processRecentBars are processed to control load:
startBar = last_bar_index - processRecentBars
activeRange = bar_index >= (startBar < 0 ? 0 : startBar)
If active, the observation is appended and the oldest one is removed:
if array.size(obsWindow) < windowLen
array.push(obsWindow, obs)
else
array.shift(obsWindow)
array.push(obsWindow, obs)
The model is ready only when the window is full.
3) Model Initialization
The script initializes a 3 state model with:
Uniform initial state probabilities
A transition matrix seeded with high persistence and equal small jump probabilities
State means initialized around zero with a configured separation
State variances initialized to a configured starting value
Key logic:
Stay probability equals initialPersistence
Jump probability equals the remaining probability split across other states
This gives the HMM a stable starting point before training.
4) Emission Model: Gaussian per State
Each state emits observations using a Gaussian density:
math.exp(-0.5 * d * d / varS) / math.sqrt(TWO_PI * varS)
Variance uses a floor:
float varS = math.max(array.get(this.vr, s), varMin)
This prevents variance collapse and numeric instability.
5) Forward Algorithm with Scaling
The script computes the forward probabilities alpha and applies scaling coefficients c to prevent underflow. It then recovers log likelihood from the scaling coefficients:
this.logLik := -sum(log(c ))
This is essential because HMM sequences quickly underflow without scaling.
6) Backward Algorithm with Scaling
The backward probabilities beta are computed using the scaling values from the forward pass, ensuring alpha and beta remain numerically stable across the entire window.
7) Expectation Step: Gamma and Xi
Gamma represents posterior probability of being in state i at time t . Xi represents posterior probability of transitioning from i to j between t and t+1 .
Xi is normalized per time step:
xij = xi_raw / denom
Gamma is computed as the sum of xi across outgoing transitions for each state:
gamma(t, i) = sum_j xi(t, i, j)
8) Maximization Step: Updating Parameters
Initial probabilities update from gamma at time 0:
pi = gamma(0, i)
Transition probabilities update from xi sums divided by gamma sums, with a transition floor and row normalization:
Each transition is clamped to transitionFloor
Each row is normalized to sum to 1
Means update as weighted averages of observations using gamma weights.
Variances update as weighted squared deviation sums with a variance floor.
9) Retraining Schedule and Online Updates
The model retrains when:
It is not initialized yet
Or the bar index matches the retrain interval
shouldRetrain = ready and (not modelInitialized or bar_index % retrainEveryBars == 0)
On retrain, Baum Welch is run for emIterations .
Between retrains, the script performs a one step forward update of the posterior:
hmm.forwardOne(posterior, obs, varianceFloor, posteriorTmp)
This provides continuous posterior updates without full retraining on every bar.
10) Role Mapping to Bull, Range, Bear
The script assigns which internal state corresponds to Bear and Bull by looking at the learned means:
Bear state is the state with the minimum mean
Bull state is the state with the maximum mean
Range is the remaining state index
This mapping updates dynamically as the model learns.
11) Regime Score and Confidence
The regime score is:
score = pBull - pBear
It is then scaled to 0 to 1:
score01 = 0.5 + 0.5 * score
Confidence is:
confidence = max(pBull, pRange, pBear)
This confidence drives background alpha and signal gating.
12) Probability Filters for Bias
Bull filter requires:
Bull probability above bullProbTrigger
Confidence above signalConfidenceMin
Bear filter requires similar conditions for Bear probability.
Bias validity adds the requirement that the dominant regime role matches the direction:
Bull bias requires dominantRole equals 1
Bear bias requires dominantRole equals minus 1
13) Extreme Pivot Logic for Reversal Candidates
The script looks for pivots in the score line:
ta.pivotlow(score01, pivotStrength, 1)
ta.pivothigh(score01, pivotStrength, 1)
It stores the most recent pivot low and pivot high along with the associated Bull or Bear probability at the pivot bar.
A low extreme is valid if:
Score at pivot is below dipScoreLevel
Bear probability at pivot exceeds extremeProbMin
A high extreme is valid if:
Score at pivot is above topScoreLevel
Bull probability at pivot exceeds extremeProbMin
14) Rebound Triggers
After an extreme, the script waits for rebound triggers:
Up rebound:
ta.crossover(score01, reboundUpLevel)
Down rebound:
ta.crossunder(score01, reboundDownLevel)
Rebound must occur within the reversal window bars from the extreme pivot.
15) Transition Edge and Asymmetry Logic
The script computes predicted probabilities of switching toward Bull or Bear using the transition matrix and current posterior. It also computes transition asymmetry between the Bull to Bear and Bear to Bull transitions.
A bullish switch condition requires:
Switch edge greater than hmmEdgeMin
Transition asymmetry favoring Bear to Bull at or above transitionAsymMin
Bull probability greater than Bear probability
A bearish switch condition uses the mirrored logic.
This adds a model based confirmation that a regime switch is plausible, not only that the score bounced.
16) Momentum Confirmation
Bull momentum requires:
Bull probability rising
Bear probability falling
Bear momentum requires the opposite.
These conditions prevent signals when probabilities are flat or conflicting.
17) Final Reversal Signal Construction
Buy reversal requires:
Valid low extreme
Not consumed
Inside reversal window
Rebound up
Bull probability and confidence filter
Bullish HMM switch condition
Bull momentum
Sell reversal requires the mirrored set of conditions.
The sell is suppressed if a buy is simultaneously true so conflicting signals do not print on the same bar.
18) Visualization Output
The script plots:
Probability curves for each regime
A dominant probability area
A thick score line colored by regime
A regime strip column plot
Fills between Bull and Bear curves and between rebound levels
Adaptive background
Optional markers for regime shifts
Reversal markers as glow plus label style plots
The info label consolidates the most important current state and threshold data for transparency. Indicator
KNN Trend Forecaster [UAlgo]KNN Trend Forecaster is a chart overlay forecasting tool that uses a K Nearest Neighbors style similarity engine to estimate the next directional bias and project a probabilistic price path. It converts the current market state into a compact feature vector, compares it to a rolling memory of historical states, and computes an expected forward change as a weighted consensus of the most similar past observations.
The indicator is built for decision support rather than signal chasing. It provides a projected path, a volatility aware tunnel around that path, and optional ghost structures that visualize how the forecast could evolve bar by bar. A minimal UI panel summarizes the current projection and sentiment, while the plot color adapts to bullish, bearish, or neutral expectation.
This script is most effective when treated as a contextual layer. It can help align trade selection with the dominant statistical bias implied by recent conditions, while still leaving execution to your own confirmation rules.
🔹 Features
1) KNN Similarity Forecasting Engine
The core model is a K Nearest Neighbors approach. For each bar, the script builds a three dimensional feature set from momentum, volatility, and relative strength. It then measures the distance between the current feature set and each stored historical feature set. The closest K neighbors are selected, and their realized forward returns are combined into a single prediction.
Model Sensitivity controls K. Lower values behave more reactive and can change bias quickly. Higher values behave more stable and tend to smooth the projection.
2) Feature Design Focused On Trading Context
The feature set is intentionally practical:
RSI captures directional pressure and mean reversion tendencies
ROC captures normalized momentum
ATR captures normalized volatility regime
Normalization ensures that ROC and ATR values are scaled into comparable ranges so the distance metric remains balanced and does not get dominated by raw magnitude differences.
3) Rolling Memory With Outcome Labels
The script builds a training memory in real time. On each confirmed bar, it stores the features from ten bars ago and labels them with the percentage change over the next ten bars. This creates a consistent supervised learning target:
Feature snapshot at time t
Outcome equals return from time t to t plus ten bars
Memory is capped to a fixed size to keep performance stable.
4) Weighted Neighbor Voting For Robust Predictions
Rather than using a simple average of neighbor outcomes, the script assigns higher weight to closer neighbors. Weight is the inverse of distance, which prioritizes highly similar historical states and reduces the influence of weaker matches.
This helps stabilize results when the market is transitioning and the feature landscape becomes noisier.
5) Forecast Path With Adaptive Step Decay
Once a prediction is produced, the script generates a forward path for a user selected Forecast Horizon. The step applied to the path decays with the square root of the forecast index, which makes the projection more confident near the present and more conservative further out.
The result is a smooth curve rather than an aggressive linear extrapolation.
6) Multi Layer Volatility Tunnel
A tunnel can be drawn around the projected path. Its width scales with ATR and expands over the forecast horizon using a square root growth profile. Tunnel Volatility controls how wide the envelope becomes.
This provides a practical view of expected dispersion around the forecast rather than a single deterministic line.
7) Ghost Structures For Bar To Bar Projection Framing
Optional ghost boxes are printed for each forward step. Each box visualizes the projected candle body from the current projected close to the next projected close. The ghost color adapts to whether the step is rising or falling, making momentum and path rhythm easier to read.
8) Neon Glow Path And Target Tag
When enabled, the path is rendered twice using polylines:
A wider glow stroke for visual depth
A thinner main stroke for precision
A target label is placed at the end of the horizon, showing the model predicted change in percent.
9) Projection Basis And Dashboard
A 21 period EMA is plotted as a reference basis, colored by the current prediction bias. A compact table displays the projection value and a sentiment label that classifies the forecast into bullish, bearish, or neutral ranges.
🔹 Calculations
1) Feature Construction
The model uses three features built from common market analytics.
RSI uses standard 14 period RSI:
float f_rsi = ta.rsi(close, 14)
ROC and ATR are normalized into a 0 to 100 style range using rolling min max scaling:
normalize(float src, int len) =>
float mn = ta.lowest(src, len)
float mx = ta.highest(src, len)
(src - mn) / (math.max(mx - mn, 0.000001)) * 100
float f_roc = normalize(ta.roc(close, 10), 100)
float f_atr = normalize(ta.atr(14), 100)
The current feature vector:
FeatureSet current_f = FeatureSet.new(f_rsi, f_roc, f_atr)
2) Training Memory And Outcome Labeling
On each confirmed bar, the script stores the feature snapshot from ten bars earlier and labels it with the forward ten bar return.
Outcome in percent:
float outcome = (close - close ) / close * 100
Training point created from the past feature snapshot:
memory.push(TrainingPoint.new(FeatureSet.new(f_rsi , f_roc , f_atr ), outcome))
Memory is limited for stability:
if memory.size() > 1000
memory.shift()
3) Distance Metric Between Feature Vectors
Similarity is computed using Euclidean distance in three dimensions:
method distance(FeatureSet v1, FeatureSet v2) =>
math.sqrt(math.pow(v1.f1 - v2.f1, 2) + math.pow(v1.f2 - v2.f2, 2) + math.pow(v1.f3 - v2.f3, 2))
Smaller distance means greater similarity.
4) Neighbor Selection And Weighted Prediction
The script computes distances from the current state to each stored training point and gathers the outcomes. It then selects the closest K entries by repeatedly taking the minimum distance.
Each neighbor is weighted by inverse distance:
float w = 1.0 / math.max(td.get(idx), 0.0001)
twc += tc.get(idx) * w
ws += w
pred := twc / ws
This produces pred, a percentage change estimate inferred from the most similar historical contexts.
5) Signal Color Classification
The display color adapts to the sign and magnitude of the prediction. Small values map to a neutral tone, stronger positive values map to bullish tone, and stronger negative values map to bearish tone.
color sig_col = pred > 0.005 ? THEME_UP : pred < -0.005 ? THEME_DN : THEME_MID
6) Forecast Path Generation
Path construction begins from the current close. A base step is derived from the prediction and then decayed across the horizon.
Base step:
float step = (pred / 10.0) * 0.01
Forward projection with square root decay:
float next_c = cur_c * (1 + step * (1.0 / math.sqrt(i)))
This produces a smooth forecast curve where early steps carry more weight than later steps.
7) Volatility Tunnel Width Model
The tunnel uses ATR as the volatility anchor and expands across the horizon:
Outer width:
float v_outer = (atr * expansion * 0.3 * math.sqrt(i))
Inner width is half of the outer width:
float v_inner = v_outer * 0.5
Upper and lower bounds are then computed around the projected close:
float h_out = next_c + v_outer
float l_out = next_c - v_outer
float h_in = next_c + v_inner
float l_in = next_c - v_inner
8) Ghost Structures
For each forecast step, a box is drawn between the current projected close and the next projected close. Its color reflects whether the path step is rising or falling.
color g_col = next_c >= cur_c ? color.new(THEME_UP, 40) : color.new(THEME_DN, 40)
box b = box.new(x1, math.max(cur_c, next_c), x2, math.min(cur_c, next_c), border_color=color.new(g_col, 20), bgcolor=g_col, border_width=1)
9) Path Rendering And Target Tag
When glow is enabled, the script renders a thick glow polyline and a thinner main polyline over the same projected points. A label is placed at the final horizon index showing the predicted percent change.
path_glow := polyline.new(pts, curved=true, line_color=color.new(sig_col, 70), line_width=8)
path_main := polyline.new(pts, curved=true, line_color=sig_col, line_width=2)
target_tag := label.new(bar_index + forecast_len, cur_c, "TARGET: " + str.tostring(pred, "#.##") + "%")
10) UI Summary And Basis Plot
A 21 period EMA is plotted and colored by the current bias. A table panel prints the projection value and a sentiment classification:
Bullish when pred is above 0.01
Bearish when pred is below minus 0.01
Neutral otherwise
This gives an at a glance readout that matches the on chart color theme. Indicator
Polynomial Regression Clustering [LuxAlgo]The Polynomial Regression Clustering indicator utilizes K-Means clustering to categorize historical price data into discrete levels and fits polynomial regression curves to each identified cluster.
This tool allows traders to visualize non-linear trends within specific price regimes, providing a unique perspective on support, resistance, and price momentum.
🔶 USAGE
The indicator identifies "K" number of clusters based on the vertical distribution of price over a user-defined lookback period. Each cluster represents a group of bars that share similar price levels, and a polynomial regression line is calculated to represent the localized trend for that specific group.
🔹 Cluster Identification
The script groups price action into color-coded dots. By default, it uses the HL2 (Average price) to determine which cluster a bar belongs to. This is particularly useful for identifying historical value areas where price has spent a significant amount of time.
🔹 Polynomial Fitting
Unlike standard linear regression, which produces a straight line, the polynomial regression curves can bend to fit the data more accurately.
A Polynomial Degree of 1 will result in a standard linear regression (straight lines). A Polynomial Degree of 2 or higher allows for curves that capture parabolic moves or cyclical swings within each cluster.
🔹 Future Projections
The current active cluster (the one containing the most recent price point) can be projected into the future. This allows you to see where the localized trend for the current price regime is heading based on the mathematical fit of historical data.
🔶 DETAILS
🔹 K-Means Algorithm
The script uses an iterative K-Means algorithm to find the optimal centroids (center points) for the price levels. It calculates the distance of each price point to the nearest centroid and refines the centroid position until the clusters are stable or the maximum iterations are reached.
🔹 Regression Logic
Once price points are assigned to a cluster, the script solves for the coefficients of a polynomial equation that minimizes the distance between the line and the cluster's data points. To ensure numerical stability with higher degrees, the horizontal (time) axis is normalized before performing matrix operations.
🔶 SETTINGS
🔹 K-Means
Number of Clusters (K): Defines how many price levels the indicator should look for. Higher values create more granular levels. Lookback Period: The number of recent bars used to perform the clustering and regression calculation. Max Iterations: The maximum number of refinement steps for the K-Means algorithm.
🔹 Regression
Polynomial Degree: Controls the "bend" of the regression lines. Higher degrees allow for more complex curves. Extend All Fits to Current Bar: When enabled, the regression lines for all historical clusters are extended to the rightmost edge of the chart. Project Current Cluster into Future: Extends the current regime's regression line into the future (empty space) using a dashed line.
🔹 Visual Style
Show Regression Lines: Toggles the visibility of the polynomial curves. Show Cluster Dots: Toggles the visibility of the colored dots on each price bar. Dot Size: Adjusts the size of the cluster dots. Cluster Colors: Customizable colors for each of the identified clusters. Indicator
Clusters Volume Profile [LuxAlgo]The Clusters Volume Profile indicator utilizes K-Means clustering to categorize historical price action into distinct groups and generates individual volume profiles for each detected cluster. This tool provides a unique perspective on volume distribution by isolating price behaviors based on proximity rather than strictly chronological order.
🔶 USAGE
The indicator identifies "clusters" of price activity within a user-defined lookback period. Each cluster is assigned a unique color and its own horizontal volume profile, allowing traders to see where liquidity is most concentrated within specific price regimes.
🔹 Identifying Institutional Zones
Traders can use the Point of Control (POC) of high-volume clusters to identify significant institutional interest. Because the K-Means algorithm groups price action by density rather than time, a cluster's POC often represents a "fair value" level where significant exchange occurred. These dashed POC lines frequently act as robust support or resistance levels when price returns to them in the future.
🔹 Market Regime Detection
By observing the vertical distribution and overlap of clusters, traders can identify market phases. Overlapping clusters with high volume often indicate accumulation or distribution phases (sideways markets), whereas distinct, vertically separated clusters with lower volume gaps between them suggest a trending environment. A shift from multiple overlapping clusters to a new, isolated cluster can signal a breakout or the start of a new trend.
🔹 Precision Entry & Exits
Cluster boundaries and POC lines provide concrete levels for trade management. An entry can be sought when price retests a high-volume cluster POC, while stops can be placed outside the total price range of that specific cluster (the area covered by its volume profile). Conversely, targets can be set at the POC of the next major cluster above or below current price action.
🔹 Volume Conviction
The tool provides specific volume metrics that allow traders to gauge conviction. By comparing the "Total" volume label of one cluster against another, a trader can determine which price regime had more participation. A breakout into a price zone with a high-volume cluster suggests stronger conviction and a higher probability of the level holding compared to a zone with low total volume.
🔶 DETAILS
The script employs a K-Means clustering algorithm. This process involves:
Initializing "centroids" across the price range of the lookback period.
Iteratively assigning each price bar to the nearest centroid based on the HLC2 (median) price.
Recalculating centroids based on the volume-weighted average price of the assigned bars.
Finalizing assignments after the specified number of iterations to ensure stable clusters.
By separating price action into these clusters, the tool helps identify high-interest zones that might be obscured by a single, traditional Volume Profile.
🔶 SETTINGS
🔹 Clustering Settings
Lookback Period: Determines the number of recent bars used for the clustering analysis.
Number of Clusters: Sets how many distinct price groups the algorithm should attempt to find (2 to 10).
K-Means Iterations: Controls the number of times the algorithm refines the cluster centers. Higher values can lead to more stable results.
🔹 Volume Profile Settings
Rows per Cluster VP: Defines the vertical resolution (number of bins) for each individual cluster's profile.
Max VP Width (Bars): Sets the maximum horizontal length of the volume profile histograms.
VP Offset: Adjusts the horizontal spacing between the current bar and the start of the volume profiles.
Highlight Price Dots: Toggles the visibility of the colored dots on the price action to identify cluster assignments.
Dot Size: Adjusts the size of the cluster assignment dots on the chart, ranging from tiny to huge.
Indicator
Machine Learning Moving Average [BackQuant]Machine Learning Moving Average
A powerful tool combining clustering, pseudo-machine learning, and adaptive prediction, enabling traders to understand and react to price behavior across multiple market regimes (Bullish, Neutral, Bearish). This script uses a dynamic clustering approach based on percentile thresholds and calculates an adaptive moving average, ideal for forecasting price movements with enhanced confidence levels.
What is Percentile Clustering?
Percentile clustering is a method that sorts and categorizes data into distinct groups based on its statistical distribution. In this script, the clustering process relies on the percentile values of a composite feature (based on technical indicators like RSI, CCI, ATR, etc.). By identifying key thresholds (lower and upper percentiles), the script assigns each data point (price movement) to a cluster (Bullish, Neutral, or Bearish), based on its proximity to these thresholds.
This approach mimics aspects of machine learning, where we “train” the model on past price behavior to predict future movements. The key difference is that this is not true machine learning; rather, it uses data-driven statistical techniques to "cluster" the market into patterns.
Why Percentile Clustering is Useful
Clustering price data into meaningful patterns (Bullish, Neutral, Bearish) helps traders visualize how price behavior can be grouped over time.
By leveraging past price behavior and technical indicators, percentile clustering adapts dynamically to evolving market conditions.
It helps you understand whether price behavior today aligns with past bullish or bearish trends, improving market context.
Clusters can be used to predict upcoming market conditions by identifying regimes with high confidence, improving entry/exit timing.
What This Script Does
Clustering Based on Percentiles : The script uses historical price data and various technical features to compute a "composite feature" for each bar. This feature is then sorted and clustered based on predefined percentile thresholds (e.g., 10th percentile for lower, 90th percentile for upper).
Cluster-Based Prediction : Once clustered, the script uses a weighted average, cluster momentum, or regime transition model to predict future price behavior over a specified number of bars.
Dynamic Moving Average : The script calculates a machine-learning-inspired moving average (MLMA) based on the current cluster, adjusting its behavior according to the cluster regime (Bullish, Neutral, Bearish).
Adaptive Confidence Levels : Confidence in the predicted return is calculated based on the distance between the current value and the other clusters. The further it is from the next closest cluster, the higher the confidence.
Visual Cluster Mapping : The script visually highlights different clusters on the chart with distinct colors for Bullish, Neutral, and Bearish regimes, and plots the MLMA line.
Prediction Output : It projects the predicted price based on the selected method and shows both predicted price and confidence percentage for each prediction horizon.
Trend Identification : Using the clustering output, the script colors the bars based on the current cluster to reflect whether the market is trending Bullish (green), Bearish (red), or is Neutral (gray).
How Traders Use It
Predicting Price Movements : The script provides traders with an idea of where prices might go based on past market behavior. Traders can use this forecast for short-term and long-term predictions, guiding their trades.
Clustering for Regime Analysis : Traders can identify whether the market is in a Bullish, Neutral, or Bearish regime, using that information to adjust trading strategies.
Adaptive Moving Average for Trend Following : The adaptive moving average can be used as a trend-following indicator, helping traders stay in the market when it’s aligned with the current trend (Bullish or Bearish).
Entry/Exit Strategy : By understanding the current cluster and its associated trend, traders can time entries and exits with higher precision, taking advantage of favorable conditions when the confidence in the predicted price is high.
Confidence for Risk Management : The confidence level associated with the predicted returns allows traders to manage risk better. Higher confidence levels indicate stronger market conditions, which can lead to higher position sizes.
Pseudo Machine Learning Aspect
While the script does not use conventional machine learning models (e.g., neural networks or decision trees), it mimics certain aspects of machine learning in its approach. By using clustering and the dynamic adjustment of a moving average, the model learns from historical data to adjust predictions for future price behavior. The "learning" comes from how the script uses past price data (and technical indicators) to create patterns (clusters) and predict future market movements based on those patterns.
Why This Is Important for Traders
Understanding market regimes helps to adjust trading strategies in a way that adapts to current market conditions.
Forecasting price behavior provides an additional edge, enabling traders to time entries and exits based on predicted price movements.
By leveraging the clustering technique, traders can separate noise from signal, improving the reliability of trading signals.
The combination of clustering and predictive modeling in one tool reduces the complexity for traders, allowing them to focus on actionable insights rather than manual analysis.
How to Interpret the Output
Bullish (Green) Zone : When the price behavior clusters into the Bullish zone, expect upward price movement. The MLMA line will help confirm if the trend remains upward.
Bearish (Red) Zone : When the price behavior clusters into the Bearish zone, expect downward price movement. The MLMA line will assist in tracking any downward trends.
Neutral (Gray) Zone : A neutral market condition signals indecision or range-bound behavior. The MLMA line can help track any potential breakouts or trend reversals.
Predicted Price : The projected price is shown on the chart, based on the cluster's predicted behavior. This provides a useful reference for where the price might move in the near future.
Prediction Confidence : The confidence percentage helps you gauge the reliability of the predicted price. A higher percentage indicates stronger market confidence in the forecasted move.
Tips for Use
Combining with Other Indicators : Use the output of this indicator in combination with your existing strategy (e.g., RSI, MACD, or moving averages) to enhance signal accuracy.
Position Sizing with Confidence : Increase position size when the prediction confidence is high, and decrease size when it’s low, based on the confidence interval.
Regime-Based Strategy : Consider developing a multi-strategy approach where you use this tool for Bullish or Bearish regimes and a separate strategy for Neutral markets.
Optimization : Adjust the lookback period and percentile settings to optimize the clustering algorithm based on your asset’s characteristics.
Conclusion
The Machine Learning Moving Average offers a novel approach to price prediction by leveraging percentile clustering and a dynamically adapting moving average. While not a traditional machine learning model, this tool mimics the adaptive behavior of machine learning by adjusting to evolving market conditions, helping traders predict price movements and identify trends with improved confidence and accuracy.
Indicator
Machine Learning BBPct [BackQuant]Machine Learning BBPct
What this is (in one line)
A Bollinger Band %B oscillator enhanced with a simplified K-Nearest Neighbors (KNN) pattern matcher. The model compares today’s context (volatility, momentum, volume, and position inside the bands) to similar situations in recent history and blends that historical consensus back into the raw %B to reduce noise and improve context awareness. It is informational and diagnostic—designed to describe market state, not to sell a trading system.
Background: %B in plain terms
Bollinger %B measures where price sits inside its dynamic envelope: 0 at the lower band, 1 at the upper band, ~ 0.5 near the basis (the moving average). Readings toward 1 indicate pressure near the envelope’s upper edge (often strength or stretch), while readings toward 0 indicate pressure near the lower edge (often weakness or stretch). Because bands adapt to volatility, %B is naturally comparable across regimes.
Why add (simplified) KNN?
Classic %B is reactive and can be whippy in fast regimes. The simplified KNN layer builds a “nearest-neighbor memory” of recent market states and asks: “When the market looked like this before, where did %B tend to be next bar?” It then blends that estimate with the current %B. Key ideas:
• Feature vector . Each bar is summarized by up to five normalized features:
– %B itself (normalized)
– Band width (volatility proxy)
– Price momentum (ROC)
– Volume momentum (ROC of volume)
– Price position within the bands
• Distance metric . Euclidean distance ranks the most similar recent bars.
• Prediction . Average the neighbors’ prior %B (lagged to avoid lookahead), inverse-weighted by distance.
• Blend . Linearly combine raw %B and KNN-predicted %B with a configurable weight; optional filtering then adapts to confidence.
This remains “simplified” KNN: no training/validation split, no KD-trees, no scaling beyond windowed min-max, and no probabilistic calibration.
How the script is organized (by input groups)
1) BBPct Settings
• Price Source – Which price to evaluate (%B is computed from this).
• Calculation Period – Lookback for SMA basis and standard deviation.
• Multiplier – Standard deviation width (e.g., 2.0).
• Apply Smoothing / Type / Length – Optional smoothing of the %B stream before ML (EMA, RMA, DEMA, TEMA, LINREG, HMA, etc.). Turning this off gives you the raw %B.
2) Thresholds
• Overbought/Oversold – Default 0.8 / 0.2 (inside ).
• Extreme OB/OS – Stricter zones (e.g., 0.95 / 0.05) to flag stretch conditions.
3) KNN Machine Learning
• Enable KNN – Switch between pure %B and hybrid.
• K (neighbors) – How many historical analogs to blend (default 8).
• Historical Period – Size of the search window for neighbors.
• ML Weight – Blend between raw %B and KNN estimate.
• Number of Features – Use 2–5 features; higher counts add context but raise the risk of overfitting in short windows.
4) Filtering
• Method – None, Adaptive, Kalman-style (first-order),
or Hull smoothing.
• Strength – How aggressively to smooth. “Adaptive” uses model confidence to modulate its alpha: higher confidence → stronger reliance on the ML estimate.
5) Performance Tracking
• Win-rate Period – Simple running score of past signal outcomes based on target/stop/time-out logic (informational, not a robust backtest).
• Early Entry Lookback – Horizon for forecasting a potential threshold cross.
• Profit Target / Stop Loss – Used only by the internal win-rate heuristic.
6) Self-Optimization
• Enable Self-Optimization – Lightweight, rolling comparison of a few canned settings (K = 8/14/21 via simple rules on %B extremes).
• Optimization Window & Stability Threshold – Governs how quickly preferred K changes and how sensitive the overfitting alarm is.
• Adaptive Thresholds – Adjust the OB/OS lines with volatility regime (ATR ratio), widening in calm markets and tightening in turbulent ones (bounded 0.7–0.9 and 0.1–0.3).
7) UI Settings
• Show Table / Zones / ML Prediction / Early Signals – Toggle informational overlays.
• Signal Line Width, Candle Painting, Colors – Visual preferences.
Step-by-step logic
A) Compute %B
Basis = SMA(source, len); dev = stdev(source, len) × multiplier; Upper/Lower = Basis ± dev.
%B = (price − Lower) / (Upper − Lower). Optional smoothing yields standardBB .
B) Build the feature vector
All features are min-max normalized over the KNN window so distances are in comparable units. Features include normalized %B, normalized band width, normalized price ROC, normalized volume ROC, and normalized position within bands. You can limit to the first N features (2–5).
C) Find nearest neighbors
For each bar inside the lookback window, compute the Euclidean distance between current features and that bar’s features. Sort by distance, keep the top K .
D) Predict and blend
Use inverse-distance weights (with a strong cap for near-zero distances) to average neighbors’ prior %B (lagged by one bar). This becomes the KNN estimate. Blend it with raw %B via the ML weight. A variance of neighbor %B around the prediction becomes an uncertainty proxy ; combined with a stability score (how long parameters remain unchanged), it forms mlConfidence ∈ . The Adaptive filter optionally transforms that confidence into a smoothing coefficient.
E) Adaptive thresholds
Volatility regime (ATR(14) divided by its 50-bar SMA) nudges OB/OS thresholds wider or narrower within fixed bounds. The aim: comparable extremeness across regimes.
F) Early entry heuristic
A tiny two-step slope/acceleration probe extrapolates finalBB forward a few bars. If it is on track to cross OB/OS soon (and slope/acceleration agree), it flags an EARLY_BUY/SELL candidate with an internal confidence score. This is explicitly a heuristic—use as an attention cue, not a signal by itself.
G) Informational win-rate
The script keeps a rolling array of trade outcomes derived from signal transitions + rudimentary exits (target/stop/time). The percentage shown is a rough diagnostic , not a validated backtest.
Outputs and visual language
• ML Bollinger %B (finalBB) – The main line after KNN blending and optional filtering.
• Gradient fill – Greenish tones above 0.5, reddish below, with intensity following distance from the midline.
• Adaptive zones – Overbought/oversold and extreme bands; shaded backgrounds appear at extremes.
• ML Prediction (dots) – The KNN estimate plotted as faint circles; becomes bright white when confidence > 0.7.
• Early arrows – Optional small triangles for approaching OB/OS.
• Candle painting – Light green above the midline, light red below (optional).
• Info panel – Current value, signal classification, ML confidence, optimized K, stability, volatility regime, adaptive thresholds, overfitting flag, early-entry status, and total signals processed.
Signal classification (informational)
The indicator does not fire trade commands; it labels state:
• STRONG_BUY / STRONG_SELL – finalBB beyond extreme OS/OB thresholds.
• BUY / SELL – finalBB beyond adaptive OS/OB.
• EARLY_BUY / EARLY_SELL – forecast suggests a near-term cross with decent internal confidence.
• NEUTRAL – between adaptive bands.
Alerts (what you can automate)
• Entering adaptive OB/OS and extreme OB/OS.
• Midline cross (0.5).
• Overfitting detected (frequent parameter flipping).
• Early signals when early confidence > 0.7.
These are purely descriptive triggers around the indicator’s state.
Practical interpretation
• Mean-reversion context – In range markets, adaptive OS/OB with ML smoothing can reduce whipsaws relative to raw %B.
• Trend context – In persistent trends, the KNN blend can keep finalBB nearer the mid/upper region during healthy pullbacks if history supports similar contexts.
• Regime awareness – Watch the volatility regime and adaptive thresholds. If thresholds compress (high vol), “OB/OS” comes sooner; if thresholds widen (calm), it takes more stretch to flag.
• Confidence as a weight – High mlConfidence implies neighbors agree; you may rely more on the ML curve. Low confidence argues for de-emphasizing ML and leaning on raw %B or other tools.
• Stability score – Rising stability indicates consistent parameter selection and fewer flips; dropping stability hints at a shifting backdrop.
Methodological notes
• Normalization uses rolling min-max over the KNN window. This is simple and scale-agnostic but sensitive to outliers; the distance metric will reflect that.
• Distance is unweighted Euclidean. If you raise featureCount, you increase dimensionality; consider keeping K larger and lookback ample to avoid sparse-neighbor artifacts.
• Lag handling intentionally uses neighbors’ previous %B for prediction to avoid lookahead bias.
• Self-optimization is deliberately modest: it only compares a few canned K/threshold choices using simple “did an extreme anticipate movement?” scoring, then enforces a stability regime and an overfitting guard. It is not a grid search or GA.
• Kalman option is a first-order recursive filter (fixed gain), not a full state-space estimator.
• Hull option derives a dynamic length from 1/strength; it is a convenience smoothing alternative.
Limitations and cautions
• Non-stationarity – Nearest neighbors from the recent window may not represent the future under structural breaks (policy shifts, liquidity shocks).
• Curse of dimensionality – Adding features without sufficient lookback can make genuine neighbors rare.
• Overfitting risk – The script includes a crude overfitting detector (frequent parameter flips) and will fall back to defaults when triggered, but this is only a guardrail.
• Win-rate display – The internal score is illustrative; it does not constitute a tradable backtest.
• Latency vs. smoothness – Smoothing and ML blending reduce noise but add lag; tune to your timeframe and objectives.
Tuning guide
• Short-term scalping – Lower len (10–14), slightly lower multiplier (1.8–2.0), small K (5–8), featureCount 3–4, Adaptive filter ON, moderate strength.
• Swing trading – len (20–30), multiplier ~2.0, K (8–14), featureCount 4–5, Adaptive thresholds ON, filter modest.
• Strong trends – Consider higher adaptive_upper/lower bounds (or let volatility regime do it), keep ML weight moderate so raw %B still reflects surges.
• Chop – Higher ML weight and stronger Adaptive filtering; accept lag in exchange for fewer false extremes.
How to use it responsibly
Treat this as a state descriptor and context filter. Pair it with your execution signals (structure breaks, volume footprints, higher-timeframe bias) and risk management. If mlConfidence is low or stability is falling, lean less on the ML line and more on raw %B or external confirmation.
Summary
Machine Learning BBPct augments a familiar oscillator with a transparent, simplified KNN memory of recent conditions. By blending neighbors’ behavior into %B and adapting thresholds to volatility regime—while exposing confidence, stability, and a plain early-entry heuristic—it provides an informational, probability-minded view of stretch and reversion that you can interpret alongside your own process. Indicator
Bober XM v2.0# ₿ober XM v2.0 Trading Bot Documentation
**Developer's Note**: While our previous Bot 1.3.1 was removed due to guideline violations, this setback only fueled our determination to create something even better. Rising from this challenge, Bober XM 2.0 emerges not just as an update, but as a complete reimagining with multi-timeframe analysis, enhanced filters, and superior adaptability. This adversity pushed us to innovate further and deliver a strategy that's smarter, more agile, and more powerful than ever before. Challenges create opportunity - welcome to Cryptobeat's finest work yet.
## !!!!You need to tune it for your own pair and timeframe and retune it periodicaly!!!!!
## Overview
The ₿ober XM v2.0 is an advanced dual-channel trading bot with multi-timeframe analysis capabilities. It integrates multiple technical indicators, customizable risk management, and advanced order execution via webhook for automated trading. The bot's distinctive feature is its separate channel systems for long and short positions, allowing for asymmetric trade strategies that adapt to different market conditions across multiple timeframes.
### Key Features
- **Multi-Timeframe Analysis**: Analyze price data across multiple timeframes simultaneously
- **Dual Channel System**: Separate parameter sets for long and short positions
- **Advanced Entry Filters**: RSI, Volatility, Volume, Bollinger Bands, and KEMAD filters
- **Machine Learning Moving Average**: Adaptive prediction-based channels
- **Multiple Entry Strategies**: Breakout, Pullback, and Mean Reversion modes
- **Risk Management**: Customizable stop-loss, take-profit, and trailing stop settings
- **Webhook Integration**: Compatible with external trading bots and platforms
### Strategy Components
| Component | Description |
|---------|-------------|
| **Dual Channel Trading** | Uses either Keltner Channels or Machine Learning Moving Average (MLMA) with separate settings for long and short positions |
| **MLMA Implementation** | Machine learning algorithm that predicts future price movements and creates adaptive bands |
| **Pivot Point SuperTrend** | Trend identification and confirmation system based on pivot points |
| **Three Entry Strategies** | Choose between Breakout, Pullback, or Mean Reversion approaches |
| **Advanced Filter System** | Multiple customizable filters with multi-timeframe support to avoid false signals |
| **Custom Exit Logic** | Exits based on OBV crossover of its moving average combined with pivot trend changes |
### Note for Novice Users
This is a fully featured real trading bot and can be tweaked for any ticker — SOL is just an example. It follows this structure:
1. **Indicator** – gives the initial signal
2. **Entry strategy** – decides when to open a trade
3. **Exit strategy** – defines when to close it
4. **Trend confirmation** – ensures the trade follows the market direction
5. **Filters** – cuts out noise and avoids weak setups
6. **Risk management** – controls losses and protects your capital
To tune it for a different pair, you'll need to start from scratch:
1. Select the timeframe (candle size)
2. Turn off all filters and trend entry/exit confirmations
3. Choose a channel type, channel source and entry strategy
4. Adjust risk parameters
5. Tune long and short settings for the channel
6. Fine-tune the Pivot Point Supertrend and Main Exit condition OBV
This will generate a lot of signals and activity on the chart. Your next task is to find the right combination of filters and settings to reduce noise and tune it for profitability.
### Default Strategy values
Default values are tuned for: Symbol BITGET:SOLUSDT.P 5min candle
Filters are off by default: Try to play with it to understand how it works
## Configuration Guide
### General Settings
| Setting | Description | Default Value |
|---------|-------------|---------------|
| **Long Positions** | Enable or disable long trades | Enabled |
| **Short Positions** | Enable or disable short trades | Enabled |
| **Risk/Reward Area** | Visual display of stop-loss and take-profit zones | Enabled |
| **Long Entry Source** | Price data used for long entry signals | hl2 (High+Low/2) |
| **Short Entry Source** | Price data used for short entry signals | hl2 (High+Low/2) |
The bot allows you to trade long positions, short positions, or both simultaneously. Each direction has its own set of parameters, allowing for fine-tuned strategies that recognize the asymmetric nature of market movements.
### Multi-Timeframe Settings
1. **Enable Multi-Timeframe Analysis**: Toggle 'Enable Multi-Timeframe Analysis' in the Multi-Timeframe Settings section
2. **Configure Timeframes**: Set appropriate higher timeframes based on your trading style:
- Timeframe 1: Default is now 15 minutes (intraday confirmation)
- Timeframe 2: Default is 4 hours (trend direction)
3. **Select Sources per Indicator**: For each indicator (RSI, KEMAD, Volume, etc.), choose:
- The desired timeframe (current, mtf1, or mtf2)
- The appropriate price type (open, high, low, close, hl2, hlc3, ohlc4)
### Entry Strategies
- **Breakout**: Enter when price breaks above/below the channel
- **Pullback**: Enter when price pulls back to the channel
- **Mean Reversion**: Enter when price is extended from the channel
You can enable different strategies for long and short positions.
### Core Components
### Risk Management
- **Position Size**: Control risk with percentage-based position sizing
- **Stop Loss Options**:
- Fixed: Set a specific price or percentage from entry
- ATR-based: Dynamic stop-loss based on market volatility
- Swing: Uses recent swing high/low points
- **Take Profit**: Multiple targets with percentage allocation
- **Trailing Stop**: Dynamic stop that follows price movement
## Advanced Usage Strategies
### Moving Average Type Selection Guide
- **SMA**: More stable in choppy markets, good for higher timeframes
- **EMA/WMA**: More responsive to recent price changes, better for entry signals
- **VWMA**: Adds volume weighting for stronger trends, use with Volume filter
- **HMA**: Balance between responsiveness and noise reduction, good for volatile markets
### Multi-Timeframe Strategy Approaches
- **Trend Confirmation**: Use higher timeframe RSI (mtf2) for overall trend, current timeframe for entries
- **Entry Precision**: Use KEMAD on current timeframe with volume filter on mtf1
- **False Signal Reduction**: Apply RSI filter on mtf1 with strict KEMAD settings
### Market Condition Optimization
| Market Condition | Recommended Settings |
|------------------|----------------------|
| **Trending** | Use Breakout strategy with KEMAD filter on higher timeframe |
| **Ranging** | Use Mean Reversion with strict RSI filter (mtf1) |
| **Volatile** | Increase ATR multipliers, use HMA for moving averages |
| **Low Volatility** | Decrease noise parameters, use pullback strategy |
## Webhook Integration
The strategy features a professional webhook system that allows direct connectivity to your exchange or trading platform of choice through third-party services like 3commas, Alertatron, or Autoview.
The webhook payload includes all necessary parameters for automated execution:
- Entry price and direction
- Stop loss and take profit levels
- Position size
- Custom identifier for webhook routing
## Performance Optimization Tips
1. **Start with Defaults**: Begin with the default settings for your timeframe before customizing
2. **Adjust One Component at a Time**: Make incremental changes and test the impact
3. **Match MA Types to Market Conditions**: Use appropriate moving average types based on the Market Condition Optimization table
4. **Timeframe Synergy**: Create logical relationships between timeframes (e.g., 5min chart with 15min and 4h higher timeframes)
5. **Periodic Retuning**: Markets evolve - regularly review and adjust parameters
## Common Setups
### Crypto Trend-Following
- MLMA with EMA or HMA
- Higher RSI thresholds (75/25)
- KEMAD filter on mtf1
- Breakout entry strategy
### Stock Swing Trading
- MLMA with SMA for stability
- Volume filter with higher threshold
- KEMAD with increased filter order
- Pullback entry strategy
### Forex Scalping
- MLMA with WMA and lower noise parameter
- RSI filter on current timeframe
- Use highest timeframe for trend direction only
- Mean Reversion strategy
## Webhook Configuration
- **Benefits**:
- Automated trade execution without manual intervention
- Immediate response to market conditions
- Consistent execution of your strategy
- **Implementation Notes**:
- Requires proper webhook configuration on your exchange or platform
- Test thoroughly with small position sizes before full deployment
- Consider latency between signal generation and execution
### Backtesting Period
Define a specific historical period to evaluate the bot's performance:
| Setting | Description | Default Value |
|---------|-------------|---------------|
| **Start Date** | Beginning of backtest period | January 1, 2025 |
| **End Date** | End of backtest period | December 31, 2026 |
- **Best Practice**: Test across different market conditions (bull markets, bear markets, sideways markets)
- **Limitation**: Past performance doesn't guarantee future results
## Entry and Exit Strategies
### Dual-Channel System
A key innovation of the Bober XM is its dual-channel approach:
- **Independent Parameters**: Each trade direction has its own channel settings
- **Asymmetric Trading**: Recognizes that markets often behave differently in uptrends versus downtrends
- **Optimized Performance**: Fine-tune settings for both bullish and bearish conditions
This approach allows the bot to adapt to the natural asymmetry of markets, where uptrends often develop gradually while downtrends can be sharp and sudden.
### Channel Types
#### 1. Keltner Channels
Traditional volatility-based channels using EMA and ATR:
| Setting | Long Default | Short Default |
|---------|--------------|---------------|
| **EMA Length** | 37 | 20 |
| **ATR Length** | 13 | 17 |
| **Multiplier** | 1.4 | 1.9 |
| **Source** | low | high |
- **Strengths**:
- Reliable in trending markets
- Less prone to whipsaws than Bollinger Bands
- Clear visual representation of volatility
- **Weaknesses**:
- Can lag during rapid market changes
- Less effective in choppy, non-trending markets
#### 2. Machine Learning Moving Average (MLMA)
Advanced predictive model using kernel regression (RBF kernel):
| Setting | Description | Options |
|---------|-------------|--------|
| **Source MA** | Price data used for MA calculations | Any price source (low/high/close/etc.) |
| **Moving Average Type** | Type of MA algorithm for calculations | SMA, EMA, WMA, VWMA, RMA, HMA |
| **Trend Source** | Price data used for trend determination | Any price source (close default) |
| **Window Size** | Historical window for MLMA calculations | 5+ (default: 16) |
| **Forecast Length** | Number of bars to forecast ahead | 1+ (default: 3) |
| **Noise Parameter** | Controls smoothness of prediction | 0.01+ (default: ~0.43) |
| **Band Multiplier** | Multiplier for channel width | 0.1+ (default: 0.5-0.6) |
- **Strengths**:
- Predictive rather than reactive
- Adapts quickly to changing market conditions
- Better at identifying trend reversals early
- **Weaknesses**:
- More computationally intensive
- Requires careful parameter tuning
- Can be sensitive to input data quality
### Entry Strategies
| Strategy | Description | Ideal Market Conditions |
|----------|-------------|-------------------------|
| **Breakout** | Enters when price breaks through channel bands, indicating strong momentum | High volatility, emerging trends |
| **Pullback** | Enters when price retraces to the middle band after testing extremes | Established trends with regular pullbacks |
| **Mean Reversion** | Enters at channel extremes, betting on a return to the mean | Range-bound or oscillating markets |
#### Breakout Strategy (Default)
- **Implementation**: Enters long when price crosses above the upper band, short when price crosses below the lower band
- **Strengths**: Captures strong momentum moves, performs well in trending markets
- **Weaknesses**: Can lead to late entries, higher risk of false breakouts
- **Optimization Tips**:
- Increase channel multiplier for fewer but more reliable signals
- Combine with volume confirmation for better accuracy
#### Pullback Strategy
- **Implementation**: Enters long when price pulls back to middle band during uptrend, short during downtrend pullbacks
- **Strengths**: Better entry prices, lower risk, higher probability setups
- **Weaknesses**: Misses some strong moves, requires clear trend identification
- **Optimization Tips**:
- Use with trend filters to confirm overall direction
- Adjust middle band calculation for market volatility
#### Mean Reversion Strategy
- **Implementation**: Enters long at lower band, short at upper band, expecting price to revert to the mean
- **Strengths**: Excellent entry prices, works well in ranging markets
- **Weaknesses**: Dangerous in strong trends, can lead to fighting the trend
- **Optimization Tips**:
- Implement strong trend filters to avoid counter-trend trades
- Use smaller position sizes due to higher risk nature
### Confirmation Indicators
#### Pivot Point SuperTrend
Combines pivot points with ATR-based SuperTrend for trend confirmation:
| Setting | Default Value |
|---------|---------------|
| **Pivot Period** | 25 |
| **ATR Factor** | 2.2 |
| **ATR Period** | 41 |
- **Function**: Identifies significant market turning points and confirms trend direction
- **Implementation**: Requires price to respect the SuperTrend line for trade confirmation
#### Weighted Moving Average (WMA)
Provides additional confirmation layer for entries:
| Setting | Default Value |
|---------|---------------|
| **Period** | 15 |
| **Source** | ohlc4 (average of Open, High, Low, Close) |
- **Function**: Confirms trend direction and filters out low-quality signals
- **Implementation**: Price must be above WMA for longs, below for shorts
### Exit Strategies
#### On-Balance Volume (OBV) Based Exits
Uses volume flow to identify potential reversals:
| Setting | Default Value |
|---------|---------------|
| **Source** | ohlc4 |
| **MA Type** | HMA (Options: SMA, EMA, WMA, RMA, VWMA, HMA) |
| **Period** | 22 |
- **Function**: Identifies divergences between price and volume to exit before reversals
- **Implementation**: Exits when OBV crosses its moving average in the opposite direction
- **Customizable MA Type**: Different MA types provide varying sensitivity to OBV changes:
- **SMA**: Traditional simple average, equal weight to all periods
- **EMA**: More weight to recent data, responds faster to price changes
- **WMA**: Weighted by recency, smoother than EMA
- **RMA**: Similar to EMA but smoother, reduces noise
- **VWMA**: Factors in volume, helpful for OBV confirmation
- **HMA**: Reduces lag while maintaining smoothness (default)
#### ADX Exit Confirmation
Uses Average Directional Index to confirm trend exhaustion:
| Setting | Default Value |
|---------|---------------|
| **ADX Threshold** | 35 |
| **ADX Smoothing** | 60 |
| **DI Length** | 60 |
- **Function**: Confirms trend weakness before exiting positions
- **Implementation**: Requires ADX to drop below threshold or DI lines to cross
## Filter System
### RSI Filter
- **Function**: Controls entries based on momentum conditions
- **Parameters**:
- Period: 15 (default)
- Overbought level: 71
- Oversold level: 23
- Multi-timeframe support: Current, MTF1 (15min), or MTF2 (4h)
- Customizable price source (open, high, low, close, hl2, hlc3, ohlc4)
- **Implementation**: Blocks long entries when RSI > overbought, short entries when RSI < oversold
### Volatility Filter
- **Function**: Prevents trading during excessive market volatility
- **Parameters**:
- Measure: ATR (Average True Range)
- Period: Customizable (default varies by timeframe)
- Threshold: Adjustable multiplier
- Multi-timeframe support
- Customizable price source
- **Implementation**: Blocks trades when current volatility exceeds threshold × average volatility
### Volume Filter
- **Function**: Ensures adequate market liquidity for trades
- **Parameters**:
- Threshold: 0.4× average (default)
- Measurement period: 5 (default)
- Moving average type: Customizable (HMA default)
- Multi-timeframe support
- Customizable price source
- **Implementation**: Requires current volume to exceed threshold × average volume
### Bollinger Bands Filter
- **Function**: Controls entries based on price relative to statistical boundaries
- **Parameters**:
- Period: Customizable
- Standard deviation multiplier: Adjustable
- Moving average type: Customizable
- Multi-timeframe support
- Customizable price source
- **Implementation**: Can require price to be within bands or breaking out of bands depending on strategy
### KEMAD Filter (Kalman EMA Distance)
- **Function**: Advanced trend confirmation using Kalman filter algorithm
- **Parameters**:
- Process Noise: 0.35 (controls smoothness)
- Measurement Noise: 24 (controls reactivity)
- Filter Order: 6 (higher = more smoothing)
- ATR Length: 8 (for bandwidth calculation)
- Upper Multiplier: 2.0 (for long signals)
- Lower Multiplier: 2.7 (for short signals)
- Multi-timeframe support
- Customizable visual indicators
- **Implementation**: Generates signals based on price position relative to Kalman-filtered EMA bands
## Risk Management System
### Position Sizing
Automatically calculates position size based on account equity and risk parameters:
| Setting | Default Value |
|---------|---------------|
| **Risk % of Equity** | 50% |
- **Implementation**:
- Position size = (Account equity × Risk %) ÷ (Entry price × Stop loss distance)
- Adjusts automatically based on volatility and stop placement
- **Best Practices**:
- Start with lower risk percentages (1-2%) until strategy is proven
- Consider reducing risk during high volatility periods
### Stop-Loss Methods
Multiple stop-loss calculation methods with separate configurations for long and short positions:
| Method | Description | Configuration |
|--------|-------------|---------------|
| **ATR-Based** | Dynamic stops based on volatility | ATR Period: 14, Multiplier: 2.0 |
| **Percentage** | Fixed percentage from entry | Long: 1.5%, Short: 1.5% |
| **PIP-Based** | Fixed currency unit distance | 10.0 pips |
- **Implementation Notes**:
- ATR-based stops adapt to changing market volatility
- Percentage stops maintain consistent risk exposure
- PIP-based stops provide precise control in stable markets
### Trailing Stops
Locks in profits by adjusting stop-loss levels as price moves favorably:
| Setting | Default Value |
|---------|---------------|
| **Stop-Loss %** | 1.5% |
| **Activation Threshold** | 2.1% |
| **Trailing Distance** | 1.4% |
- **Implementation**:
- Initial stop remains fixed until profit reaches activation threshold
- Once activated, stop follows price at specified distance
- Locks in profit while allowing room for normal price fluctuations
### Risk-Reward Parameters
Defines the relationship between risk and potential reward:
| Setting | Default Value |
|---------|---------------|
| **Risk-Reward Ratio** | 1.4 |
| **Take Profit %** | 2.4% |
| **Stop-Loss %** | 1.5% |
- **Implementation**:
- Take profit distance = Stop loss distance × Risk-reward ratio
- Higher ratios require fewer winning trades for profitability
- Lower ratios increase win rate but reduce average profit
### Filter Combinations
The strategy allows for simultaneous application of multiple filters:
- **Recommended Combinations**:
- Trending markets: RSI + KEMAD filters
- Ranging markets: Bollinger Bands + Volatility filters
- All markets: Volume filter as minimum requirement
- **Performance Impact**:
- Each additional filter reduces the number of trades
- Quality of remaining trades typically improves
- Optimal combination depends on market conditions and timeframe
### Multi-Timeframe Filter Applications
| Filter Type | Current Timeframe | MTF1 (15min) | MTF2 (4h) |
|-------------|-------------------|-------------|------------|
| RSI | Quick entries/exits | Intraday trend | Overall trend |
| Volume | Immediate liquidity | Sustained support | Market participation |
| Volatility | Entry timing | Short-term risk | Regime changes |
| KEMAD | Precise signals | Trend confirmation | Major reversals |
## Visual Indicators and Chart Analysis
The bot provides comprehensive visual feedback on the chart:
- **Channel Bands**: Keltner or MLMA bands showing potential support/resistance
- **Pivot SuperTrend**: Colored line showing trend direction and potential reversal points
- **Entry/Exit Markers**: Annotations showing actual trade entries and exits
- **Risk/Reward Zones**: Visual representation of stop-loss and take-profit levels
These visual elements allow for:
- Real-time strategy assessment
- Post-trade analysis and optimization
- Educational understanding of the strategy logic
## Implementation Guide
### PulseWire Setup
1. Load the script in PulseWire Pine Editor
2. Apply to your preferred chart and timeframe
3. Adjust parameters based on your trading preferences
4. Enable alerts for webhook integration
### Webhook Integration
1. Configure webhook URL in PulseWire alerts
2. Set up receiving endpoint on your trading platform
3. Define message format matching the bot's output
4. Test with small position sizes before full deployment
### Optimization Process
1. Backtest across different market conditions
2. Identify parameter sensitivity through multiple tests
3. Focus on risk management parameters first
4. Fine-tune entry/exit conditions based on performance metrics
5. Validate with out-of-sample testing
## Performance Considerations
### Strengths
- Adaptability to different market conditions through dual channels
- Multiple layers of confirmation reducing false signals
- Comprehensive risk management protecting capital
- Machine learning integration for predictive edge
### Limitations
- Complex parameter set requiring careful optimization
- Potential over-optimization risk with so many variables
- Computational intensity of MLMA calculations
- Dependency on proper webhook configuration for execution
### Best Practices
- Start with conservative risk settings (1-2% of equity)
- Test thoroughly in demo environment before live trading
- Monitor performance regularly and adjust parameters
- Consider market regime changes when evaluating results
## Conclusion
The ₿ober XM v2.0 represents a significant evolution in trading strategy design, combining traditional technical analysis with machine learning elements and multi-timeframe analysis. The core strength of this system lies in its adaptability and recognition of market asymmetry.
### Market Asymmetry and Adaptive Approach
The strategy acknowledges a fundamental truth about markets: bullish and bearish phases behave differently and should be treated as distinct environments. The dual-channel system with separate parameters for long and short positions directly addresses this asymmetry, allowing for optimized performance regardless of market direction.
### Targeted Backtesting Philosophy
It's counterproductive to run backtests over excessively long periods. Markets evolve continuously, and strategies that worked in previous market regimes may be ineffective in current conditions. Instead:
- Test specific market phases separately (bull markets, bear markets, range-bound periods)
- Regularly re-optimize parameters as market conditions change
- Focus on recent performance with higher weight than historical results
- Test across multiple timeframes to ensure robustness
### Multi-Timeframe Analysis as a Game-Changer
The integration of multi-timeframe analysis fundamentally transforms the strategy's effectiveness:
- **Increased Safety**: Higher timeframe confirmations reduce false signals and improve trade quality
- **Context Awareness**: Decisions made with awareness of larger trends reduce adverse entries
- **Adaptable Precision**: Apply strict filters on lower timeframes while maintaining awareness of broader conditions
- **Reduced Noise**: Higher timeframe data naturally filters market noise that can trigger poor entries
The ₿ober XM v2.0 provides traders with a framework that acknowledges market complexity while offering practical tools to navigate it. With proper setup, realistic expectations, and attention to changing market conditions, it delivers a sophisticated approach to systematic trading that can be continuously refined and optimized.
Strategy
Indicator
AI Adaptive Oscillator [PhenLabs]📊 Algorithmic Adaptive Oscillator
Version: PineScript™ v6
📌 Description
The AI Adaptive Oscillator is a sophisticated technical indicator that employs ensemble learning and adaptive weighting techniques to analyze market conditions. This innovative oscillator combines multiple traditional technical indicators through an AI-driven approach that continuously evaluates and adjusts component weights based on historical performance. By integrating statistical modeling with machine learning principles, the indicator adapts to changing market dynamics, providing traders with a responsive and reliable tool for market analysis.
🚀 Points of Innovation:
Ensemble learning framework with adaptive component weighting
Performance-based scoring system using directional accuracy
Dynamic volatility-adjusted smoothing mechanism
Intelligent signal filtering with cooldown and magnitude requirements
Signal confidence levels based on multi-factor analysis
🔧 Core Components
Ensemble Framework : Combines up to five technical indicators with performance-weighted integration
Adaptive Weighting : Continuous performance evaluation with automated weight adjustment
Volatility-Based Smoothing : Adapts sensitivity based on current market volatility
Pattern Recognition : Identifies potential reversal patterns with signal qualification criteria
Dynamic Visualization : Professional color schemes with gradient intensity representation
Signal Confidence : Three-tiered confidence assessment for trading signals
🔥 Key Features
The indicator provides comprehensive market analysis through:
Multi-Component Ensemble : Integrates RSI, CCI, Stochastic, MACD, and Volume-weighted momentum
Performance Scoring : Evaluates each component based on directional prediction accuracy
Adaptive Smoothing : Automatically adjusts based on market volatility
Pattern Detection : Identifies potential reversal patterns in overbought/oversold conditions
Signal Filtering : Prevents excessive signals through cooldown periods and minimum change requirements
Confidence Assessment : Displays signal strength through intuitive confidence indicators (average, above average, excellent)
🎨 Visualization
Gradient-Filled Oscillator : Color intensity reflects strength of market movement
Clear Signal Markers : Distinct bullish and bearish pattern signals with confidence indicators
Range Visualization : Clean representation of oscillator values from -6 to 6
Zero Line : Clear demarcation between bullish and bearish territory
Customizable Colors : Color schemes that can be adjusted to match your chart style
Confidence Symbols : Intuitive display of signal confidence (no symbol, +, or ++) alongside direction markers
📖 Usage Guidelines
⚙️ Settings Guide
Color Settings
Bullish Color
Default: #2b62fa (Blue)
This setting controls the color representation for bullish movements in the oscillator. The color appears when the oscillator value is positive (above zero), with intensity indicating the strength of the bullish momentum. A brighter shade indicates stronger bullish pressure.
Bearish Color
Default: #ce9851 (Amber)
This setting determines the color representation for bearish movements in the oscillator. The color appears when the oscillator value is negative (below zero), with intensity reflecting the strength of the bearish momentum. A more saturated shade indicates stronger bearish pressure.
Signal Settings
Signal Cooldown (bars)
Default: 10
Range: 1-50
This parameter sets the minimum number of bars that must pass before a new signal of the same type can be generated. Higher values reduce signal frequency and help prevent overtrading during choppy market conditions. Lower values increase signal sensitivity but may generate more false positives.
Min Change For New Signal
Default: 1.5
Range: 0.5-3.0
This setting defines the minimum required change in oscillator value between consecutive signals of the same type. It ensures that new signals represent meaningful changes in market conditions rather than minor fluctuations. Higher values produce fewer but potentially higher-quality signals, while lower values increase signal frequency.
AI Core Settings
Base Length
Default: 14
Minimum: 2
This fundamental setting determines the primary calculation period for all technical components in the ensemble (RSI, CCI, Stochastic, etc.). It represents the lookback window for each component’s base calculation. Shorter periods create a more responsive but potentially noisier oscillator, while longer periods produce smoother signals with potential lag.
Adaptive Speed
Default: 0.1
Range: 0.01-0.3
Controls how quickly the oscillator adapts to new market conditions through its volatility-adjusted smoothing mechanism. Higher values make the oscillator more responsive to recent price action but potentially more erratic. Lower values create smoother transitions but may lag during rapid market changes. This parameter directly influences the indicator’s adaptiveness to market volatility.
Learning Lookback Period
Default: 150
Minimum: 10
Determines the historical data range used to evaluate each ensemble component’s performance and calculate adaptive weights. This setting controls how far back the AI “learns” from past performance to optimize current signals. Longer periods provide more stable weight distribution but may be slower to adapt to regime changes. Shorter periods adapt more quickly but may overreact to recent anomalies.
Ensemble Size
Default: 5
Range: 2-5
Specifies how many technical components to include in the ensemble calculation.
Understanding The Interaction Between Settings
Base Length and Learning Lookback : The base length determines the reactivity of individual components, while the lookback period determines how their weights are adjusted. These should be balanced according to your timeframe - shorter timeframes benefit from shorter base lengths, while the lookback should generally be 10-15 times the base length for optimal learning.
Adaptive Speed and Signal Cooldown : These settings control sensitivity from different angles. Increasing adaptive speed makes the oscillator more responsive, while reducing signal cooldown increases signal frequency. For conservative trading, keep adaptive speed low and cooldown high; for aggressive trading, do the opposite.
Ensemble Size and Min Change : Larger ensembles provide more stable signals, allowing for a lower minimum change threshold. Smaller ensembles might benefit from a higher threshold to filter out noise.
Understanding Signal Confidence Levels
The indicator provides three distinct confidence levels for both bullish and bearish signals:
Average Confidence (▲ or ▼) : Basic signal that meets the minimum pattern and filtering criteria. These signals indicate potential reversals but with moderate confidence in the prediction. Consider using these as initial alerts that may require additional confirmation.
Above Average Confidence (▲+ or ▼+) : Higher reliability signal with stronger underlying metrics. These signals demonstrate greater consensus among the ensemble components and/or stronger historical performance. They offer increased probability of successful reversals and can be traded with less additional confirmation.
Excellent Confidence (▲++ or ▼++) : Highest quality signals with exceptional underlying metrics. These signals show strong agreement across oscillator components, excellent historical performance, and optimal signal strength. These represent the indicator’s highest conviction trade opportunities and can be prioritized in your trading decisions.
Confidence assessment is calculated through a multi-factor analysis including:
Historical performance of ensemble components
Degree of agreement between different oscillator components
Relative strength of the signal compared to historical thresholds
✅ Best Use Cases:
Identify potential market reversals through oscillator extremes
Filter trade signals based on AI-evaluated component weights
Monitor changing market conditions through oscillator direction and intensity
Confirm trade signals from other indicators with adaptive ensemble validation
Detect early momentum shifts through pattern recognition
Prioritize trading opportunities based on signal confidence levels
Adjust position sizing according to signal confidence (larger for ++ signals, smaller for standard signals)
⚠️ Limitations
Requires sufficient historical data for accurate performance scoring
Ensemble weights may lag during dramatic market condition changes
Higher ensemble sizes require more computational resources
Performance evaluation quality depends on the learning lookback period length
Even high confidence signals should be considered within broader market context
💡 What Makes This Unique
Adaptive Intelligence : Continuously adjusts component weights based on actual performance
Ensemble Methodology : Combines strength of multiple indicators while minimizing individual weaknesses
Volatility-Adjusted Smoothing : Provides appropriate sensitivity across different market conditions
Performance-Based Learning : Utilizes historical accuracy to improve future predictions
Intelligent Signal Filtering : Reduces noise and false signals through sophisticated filtering criteria
Multi-Level Confidence Assessment : Delivers nuanced signal quality information for optimized trading decisions
🔬 How It Works
The indicator processes market data through five main components:
Ensemble Component Calculation :
Normalizes traditional indicators to consistent scale
Includes RSI, CCI, Stochastic, MACD, and volume components
Adapts based on the selected ensemble size
Performance Evaluation :
Analyzes directional accuracy of each component
Calculates continuous performance scores
Determines adaptive component weights
Oscillator Integration :
Combines weighted components into unified oscillator
Applies volatility-based adaptive smoothing
Scales final values to -6 to 6 range
Signal Generation :
Detects potential reversal patterns
Applies cooldown and magnitude filters
Generates clear visual markers for qualified signals
Confidence Assessment :
Evaluates component agreement, historical accuracy, and signal strength
Classifies signals into three confidence tiers (average, above average, excellent)
Displays intuitive confidence indicators (no symbol, +, ++) alongside direction markers
💡 Note:
The AI Adaptive Oscillator performs optimally when used with appropriate timeframe selection and complementary indicators. Its adaptive nature makes it particularly valuable during changing market conditions, where traditional fixed-weight indicators often lose effectiveness. The ensemble approach provides a more robust analysis by leveraging the collective intelligence of multiple technical methodologies. Pay special attention to the signal confidence indicators to optimize your trading decisions - excellent (++) signals often represent the most reliable trade opportunities. Indicator
Indicator
VWAP Bands with ML [CryptoSea]VWAP Machine Learning Bands is an advanced indicator designed to enhance trading analysis by integrating VWAP with a machine learning-inspired adaptive smoothing approach. This tool helps traders identify trend-based support and resistance zones, predict potential price movements, and generate dynamic trade signals.
Key Features
Adaptive ML VWAP Calculation: Uses a dynamically adjusted SMA-based VWAP model with volatility sensitivity for improved trend analysis.
Forecasting Mechanism: The 'Forecast' parameter shifts the ML output forward, providing predictive insights into potential price movements.
Volatility-Based Band Adjustments: The 'Sigma' parameter fine-tunes the impact of volatility on ML smoothing, adapting to market conditions.
Multi-Tier Standard Deviation Bands: Includes two levels of bands to define potential breakout or mean-reversion zones.
Dynamic Trend-Based Colouring: The VWAP and ML lines change colour based on their relative positions, visually indicating bullish and bearish conditions.
Custom Signal Detection Modes: Allows traders to choose between signals from Band 1, Band 2, or both, for more tailored trade setups.
In the image below, you can see an example of the bands on higher timeframe showing good mean reversion signal opportunities, these tend to work better in ranging markets rather than strong trending ones.
How It Works
VWAP & ML Integration: The script computes VWAP and applies a machine learning-inspired adjustment using SMA smoothing and volatility-based adaptation.
Forecasting Impact: The 'Forecast' setting shifts the ML output forward in time, allowing for anticipatory trend analysis.
Volatility Scaling (Sigma): Adjusts the ML smoothing sensitivity based on market volatility, providing more responsive or stable trend lines.
Trend Confirmation via Colouring: The VWAP line dynamically switches colour depending on whether it is above or below the ML output.
Multi-Level Band Analysis: Two standard deviation-based bands provide a framework for identifying breakouts, trend reversals, or continuation patterns.
In the example below, we can see some of the most reliable signals where we have mean reversion signals from the band whilst the price is also pulling back into the VWAP, these signals have the additional confluence which can give you a higher probabilty move.
Alerts
Bullish Signal Band 1: Alerts when the price crosses above the lower ML Band 1.
Bearish Signal Band 1: Alerts when the price crosses below the upper ML Band 1.
Bullish Signal Band 2: Alerts when the price crosses above the lower ML Band 2.
Bearish Signal Band 2: Alerts when the price crosses below the upper ML Band 2.
Filtered Bullish Signal: Alerts when a bullish signal is triggered based on the selected signal detection mode.
Filtered Bearish Signal: Alerts when a bearish signal is triggered based on the selected signal detection mode.
Application
Trend & Momentum Analysis: Helps traders identify key market trends and potential momentum shifts.
Dynamic Support & Resistance: Standard deviation bands serve as adaptive price zones for potential breakouts or reversals.
Enhanced Trade Signal Confirmation: The integration of ML smoothing with VWAP provides clearer entry and exit signals.
Customizable Risk Management: Allows users to adjust parameters for fine-tuned signal detection, aligning with their trading strategy.
The VWAP Machine Learning Bands indicator offers traders an innovative tool to improve market entries, recognize potential reversals, and enhance trend analysis with intelligent data-driven signals. Indicator
Machine Learning Moving Average [LuxAlgo]The Machine Learning Moving Average (MLMA) is a responsive moving average making use of the weighting function obtained Gaussian Process Regression method. Characteristic such as responsiveness and smoothness can be adjusted by the user from the settings.
The moving average also includes bands, used to highlight possible reversals.
🔶 USAGE
The Machine Learning Moving Average smooths out noisy variations from the price, directly estimating the underlying trend in the price.
A higher "Window" setting will return a longer-term moving average while increasing the "Forecast" setting will affect the responsiveness and smoothness of the moving average, with higher positive values returning a more responsive moving average and negative values returning a smoother but less responsive moving average.
Do note that an excessively high "Forecast" setting will result in overshoots, with the moving average having a poor fit with the price.
The moving average color is determined according to the estimated trend direction based on the bands described below, shifting to blue (default) in an uptrend and fushia (default) in downtrends.
The upper and lower extremities represent the range within which price movements likely fluctuate.
Signals are generated when the price crosses above or below the band extremities, with turning points being highlighted by colored circles on the chart.
🔶 SETTINGS
Window: Calculation period of the moving average. Higher values yield a smoother average, emphasizing long-term trends and filtering out short-term fluctuations.
Forecast: Sets the projection horizon for Gaussian Process Regression. Higher values create a more responsive moving average but will result in more overshoots, potentially worsening the fit with the price. Negative values will result in a smoother moving average.
Sigma: Controls the standard deviation of the Gaussian kernel, influencing weight distribution. Higher Sigma values return a longer-term moving average.
Multiplicative Factor: Adjusts the upper and lower extremity bounds, with higher values widening the bands and lowering the amount of returned turning points.
🔶 RELATED SCRIPTS
Machine-Learning-Gaussian-Process-Regression
SuperTrend-AI-Clustering
Indicator
