Forecasting indicators - page 28

 

Neuro trend neurotrend.rar

Files:
neurotrend.rar  21 kb
nn.gif  22 kb
 

Lagrange polynomial interpolation

interpolation.mq4

 

Hi I am just testing all this indicators based on BPNN library (MA_predictor and similar) and noticed something what could be interesting.

Every time when you modify indicator but not making any changes the output is different, the future line is different, and it is changed in a cycle.

Changes are extremal.

Tested off line so the data are the same.

Does it mean something to coders in the metter of a bug has to be solved?

 
Marasek:
Hi I am just testing all this indicators based on BPNN library (MA_predictor and similar) and noticed something what could be interesting.

Every time when you modify indicator but not making any changes the output is different, the future line is different, and it is changed in a cycle.

Changes are extremal.

Tested off line so the data are the same.

Does it mean something to coders in the metter of a bug has to be solved?

It is not a bug. All predicting code will work like that

 

So, if BPNN library calculating the same data 6 times in the same moment gives 6 different results, it is absolutely usless.

Abyone has an indicator based on FANN2 library or tested it ?

 

Extrapolator

The indicator is based on several methods that can be chosen by the Method input variable :

Method 1: Fourier's extrapolation; the frequencies are calculated using the Quinn-Fernandes Algorithm

Method 2: Autocorrelation Method

Method 3: Weighted Burg Method

Method 4: Burg Method with Helme-Nikias weighting function

Method 5: Itakura-Saito (geometric) method

Method 6: Modified covariance method

Posted in Something interesting please post here

 
Tsar:
Extrapolator

The indicator is based on several methods that can be chosen by the Method input variable :

Method 1: Fourier's extrapolation; the frequencies are calculated using the Quinn-Fernandes Algorithm

Method 2: Autocorrelation Method

Method 3: Weighted Burg Method

Method 4: Burg Method with Helme-Nikias weighting function

Method 5: Itakura-Saito (geometric) method

Method 6: Modified covariance method

Posted in Something interesting please post here

yeah... but is any of it accurate enough to trade with?

 
dagoods:
yeah... but is any of it accurate enough to trade with?

dagoods

The answer has two parts:

1. It is written by one of the best coders out there (gpwr) and the extrapolation methods he included are coded flawlessly

2. Extrapolation is never a "one way street". The rule of thumb is the further in the future the extrapolation extends, the less accurate it is. Just one example : a simple ema extrapolation is estimated to have over 90% accuracy on the first future bar, but already the second bar is less than 70% and and so on. So if you plan to use it use as little future bars values as possible

It will not make wonders, but any extrapolation is a sound mathematical model and is far better than some repainting indicators that are floating around as "wonders"

 
mladen:
dagoods

The answer has two parts:

1. It is written by one of the best coders out there (gpwr) and the extrapolation methods he included are coded flawlessly

2. Extrapolation is never a "one way street". The rule of thumb is the further in the future the extrapolation extends, the less accurate it is. Just one example : a simple ema extrapolation is estimated to have over 90% accuracy on the first future bar, but already the second bar is less than 70% and and so on. So if you plan to use it use as little future bars values as possible

It will not make wonders, but any extrapolation is a sound mathematical model and is far better than some repainting indicators that are floating around as "wonders"

Good explanation

I was always wondering about the accuracy of extrapolation.

Where can I find more information about ema extrapolation?

 
TamFX:
Good explanation

I was always wondering about the accuracy of extrapolation.

Where can I find more information about ema extrapolation?

The study about ema extrapolation was published here : Journal of Forecasting - All Issues - Wiley Online Library (it is a commercial publication and that is why I can not post it here)

Reason: