[RS]Shadows Of Past, Present and FutureExperimental:
session projection into the future..
Past, Present, Future
Tell me, tell me, smiling child,
What the past is like to thee ?
'An Autumn evening soft and mild
With a wind that sighs mournfully.’
Tell me, what is the present hour ?
'A green and flowery spray
Where a young bird sits gathering its power
To mount and fly away.’
And what is the future, happy one ?
'A sea beneath a cloudless sun ;
A mighty, glorious, dazzling sea
Stretching into infinity.’
by Emily Brontë
Indicator
Indicator
Indicator
Indicator
Indicator
Indicator
Indicator
Indicator
Indicator
Indicator
Indicator
Indicator
Decaying Rate of Change Non Linear FilterThis is a potential solution to dealing with the inherent lag in most filters especially with instruments such as BTC and the effects of long periods of low volatility followed by massive volatility spikes as well as whipsaws/barts etc.
We can try and solve these issues in a number of ways, adaptive lengths, dynamic weighting etc. This filter uses a non linear weighting combined with an exponential decay rate.
With the non linear weighting the filter can become very responsive to sudden volatility spikes. We can use a short length absolute rate of change as a method to improve weighting of relative high volatility.
c1 = abs(close - close ) / close
Which gives us a fairly simple filter :
filter = sum(c1 * close,periods) / sum(c1,periods)
At this point if we want to control the relative magnitude of the ROC coefficients we can do so by raising it to a power.
c2 = pow(c1, x)
Where x approaches zero the coefficient approaches 1 or a linear filter. At x = 1 we have an unmodified coefficient and higher values increase the relative magnitude of the response. As an extreme example with x = 10 we effectively isolate the highest ROC candle within the window (which has some novel support resistance horizontals as those closes are often important). This controls the degree of responsiveness, so we can magnify the responsiveness, but with the trade off of overshoot/persistence.
So now we have the problem whereby that a highly weighted data point from a high volatility event persists within the filter window. And to a possibly extreme degree, if a reversal occurs we get a potentially large "overshoot" and in a way actually induced a large amount of lag for future price action.
This filter compensates for this effect by exponentially decaying the abs(ROC) coefficient over time, so as a high volatility event passes through the filter window it receives exponentially less weighting allowing more recent prices to receive a higher relative weighting than they would have.
c3 = c2 * pow(1 - percent_decay, periods_back)
This is somewhat similar to an EMA, however with an EMA being recursive that event will persist forever (to some degree) in the calculation. Here we are using a fixed window, so once the event is behind the window it's completely removed from the calculation
I've added Ehler's Super Smoother as an optional smoothing function as some highly non linear settings benefit from smoothing. I can't remember where I got the original SS code snippet, so if you recognize it as yours msg me and I'll link you here.
Indicator
Indicator
NAND PerceptronExperimental NAND Perceptron based upon Python template that aims to predict NAND Gate Outputs. A Perceptron is one of the foundational building blocks of nearly all advanced Neural Network layers and models for Algo trading and Machine Learning.
The goal behind this script was threefold:
To prove and demonstrate that an ACTUAL working neural net can be implemented in Pine, even if incomplete.
To pave the way for other traders and coders to iterate on this script and push the boundaries of Tradingview strategies and indicators.
To see if a self-contained neural network component for parameter optimization within Pinescript was hypothetically possible.
NOTE: This is a highly experimental proof of concept - this is NOT a ready-made template to include or integrate into existing strategies and indicators, yet (emphasis YET - neural networks have a lot of potential utility and potential when utilized and implemented properly).
Hardcoded NAND Gate outputs with Bias column (X0):
// NAND Gate + X0 Bias and Y-true
// X0 // X1 // X2 // Y
// 1 // 0 // 0 // 1
// 1 // 0 // 1 // 1
// 1 // 1 // 0 // 1
// 1 // 1 // 1 // 0
Column X0 is bias feature/input
Column X1 and X2 are the NAND Gate
Column Y is the y-true values for the NAND gate
yhat is the prediction at that timestep
F0,F1,F2,F3 are the Dot products of the Weights (W0,W1,W2) and the input features (X0,X1,X2)
Learning rate and activation function threshold are enabled by default as input parameters
Uncomment sections for more training iterations/epochs:
Loop optimizations would be amazing to have for a selectable length for training iterations/epochs but I'm not sure if it's possible in Pine with how this script is structured.
Error metrics and loss have not been implemented due to difficulty with script length and iterations vs epochs - I haven't been able to configure the input parameters to successfully predict the right values for all four y-true values for the NAND gate (only been able to get 3/4; If you're able to get all four predictions to be correct, let me know, please).
// //---- REFERENCE for final output
// A3 := 1, y0 true
// B3 := 1, y1 true
// C3 := 1, y2 true
// D3 := 0, y3 true
PLEASE READ: Source article/template and main code reference:
towardsdatascience.com
towardsdatascience.com
towardsdatascience.com Indicator
Indicator
Indicator
Indicator
Indicator
Indicator
Indicator
Indicator
Indicator
