PNN for metatrader using Parzen window classification version 2 with kernel smoothing of inputs

This is the PNN using Parzen window classification version 2 with kernel smoothing of inputs. You need to have the PFE indicator.

Here you will file the standard PNN called version 2. And the PNN with money management and kernel smoothing of inputs.

Nevertheless this is a beta version, because I am not sure that the integration of the kernel smoothing with the PNN code is correctly executed.

So this is what we have for now. Anyway the idea is interesting.

In the normal PNN Eric we have as input the difference between:


b[bar]= Close[bar]-Open[bar+AmoutOfForecastBars-1])

The idea was to replace this piece of code with something else and to see how it would affect the EA.

So I replaced this by the kernel smoothing of PFE.


b[bar] = kernel()

So this is the idea. However the number of bars of parameters is still an important parameter because it affects the initial period for training:


TrainingStartTime = Time[0] + Period()*60*AmoutOfForecastBars

Of course we can replace this piece of code by Numbars_for_Training as a parameter, but I chose not to touch this for the moment.


TrainingStartTime = Time[0] + Period()*Numbars_for_Training

Any wise look in the code would be welcome as there is not many native mql code for neural implementations.

DOWNLOAD from here

Practical limitations in the Neural net extrapolation

neural net, extrapolation

I just would like to point out something very important about the application of the principle of uncertainty the technical analysis and the Elliotware approach in particular. Some may think that if they can use advanced machine learning solutions they can totally not take into account the common technical analysis. I think it is true if the model is base on tic data and high frequency trading there are other rules and principles. However when we go higher the technical analysis becomes important. But what I mean. I mean precisely the support and resustence zones.

Those zones are known as decision zones. That means that there a decision has to be made.

And this is analogy with the principle of uncertainly (a quote from wikipedia).


In quantum mechanics, the Heisenberg uncertainty principle states a fundamental limit on the accuracy with which certain pairs of physical properties of a particle, such as position and momentum, cannot be simultaneously known. In other words, the more precisely one property is measured, the less precisely the other can be controlled, determined, or known.

Well this is just an analogy. In technical analysis when we are closer to such a decision zone. The scenarios are clearly measured and predetermined. The price is either doing this or that. But if we know precusely the scenarios we do not know the actual direction.

And vice-versa when we are away from those decision zones. We may know very well the general direction of the market but we have not a clue of the trajectories.

And in that particular situation the Neural Net by their extrapolating abilities are giving us a help. When the direction is clear but the scenarios are unclear.

If you are clearly near a decision zone the performance of the neural net will depend more on luck than other things.

Here I woul like to post my shot. Here we have two scenarios. Those scenarios are clearly cut. One is going high, the other is goind down to the channel. From the downside of the channel a new decision has to be made.

I used BPNN Caterpillar with the same training period on 30 m time frame. Both of the BPNN was with lag bars 40. However one had 3 computations the other 5. One of the nets was pointing high the other low.

I mean a slight change of the inputs and we have a totally different prediction. I think if we consider a decision zone it is normal. Basically what inputs will you use and the outcome is based on luck.

BPNN Predictor plus

There is a library with mods for the BPNN Predictor Plus.The library allow to use different inputs for the neural net.

beathespread.com

The BPNN Predictor uses for training algorithm the

Improved Resilient back-Propagation Plus (iRProp+).

The method is described on this address:
http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.17.1332

The idea is to use as inputs different averaging algorithms. One of the mods includes as inputs the Singular Spectrum Analysis. In fact I use two types the normal one and the end - pointed version. 

You can choose from many different averaging algorithms.

List of MAs:

MA_Method= 0: SMA - Simple Moving Average
MA_Method= 1: EMA - Exponential Moving Average
MA_Method= 2: Wilder - Wilder Exponential Moving Average
MA_Method= 3: LWMA - Linear Weighted Moving Average 
MA_Method= 4: SineWMA - Sine Weighted Moving Average
MA_Method= 5: TriMA - Triangular Moving Average
MA_Method= 6: LSMA - Least Square Moving Average (or EPMA, Linear Regression Line)
MA_Method= 7: SMMA - Smoothed Moving Average
MA_Method= 8: HMA - Hull Moving Average by Alan Hull
MA_Method= 9: ZeroLagEMA - Zero-Lag Exponential Moving Average
MA_Method=10: DEMA - Double Exponential Moving Average by Patrick Mulloy
MA_Method=11: T3 - T3 by T.Tillson
MA_Method=12: ITrend - Instantaneous Trendline by J.Ehlers
MA_Method=13: Median - Moving Median
MA_Method=14: GeoMean - Geometric Mean
MA_Method=15: REMA - Regularized EMA by Chris Satchwell
MA_Method=16: ILRS - Integral of Linear Regression Slope 
MA_Method=17: IE/2 - Combination of LSMA and ILRS 
MA_Method=18: TriMAgen - Triangular Moving Average generalized by J.Ehlers
MA_Method=19: VWMA - Volume Weighted Moving Average 
MA_Method=20: JSmooth - Smoothing by Mark Jurik

Anyway you may see that the predictions varies a lot and it changes from time. There is nothing you can do about that. In fact the Forex time series have a lot of local optima and the neural net may constantly bounce from local optimum solution to another. 

Elliotware: How it works?

This is how Elliotware works:

1. I identify an Elliot Wave structure. This is MANUAL we do not use software. And this is the fundamental difference between this approach and the other approaches. They try to use a machine to force an Elliott Wave analysis.

2. I set the training range to the Elliot wave structure.

3. I apply my machine learning algorithm:  And I have a prediction.