The Extrapolator Indicator

The Extrapolator Indicator

In the basis of this forex - indicator a few methods which can get out to the entrance variable Method are fixed:
1: Extrapolation of the Fure row; frequencies are calculated using Quinn-Fernandes Algorithm
2: Autocorrelation Method
3: Weighted Burg Method
4: Burg Method with Helme-Nikias weighting function
5: Itakura-Saito (geometric) method
6: Modified covariance method

The methods 2-6 are the methods of linear prediction (linear prediction). The linear prediction is based on finding of future values as linear functions of past values. It is possible is present row of costs of x[0]..x[n-1] where more senior index corresponds to more recent prices. The prediction of future cost of x[n] is as
[n] = -Sum(and[I]*x[n-i], I=1..p)

where and[I=1..p] - model coefficients, p - model order. The transferred methods 2-6 find the coefficients a[] by reduction of middle-quadratic error on training n-p the last bars. It is certainly possible to attain the zero error of prediction on training bars if to decide the system of linear equalizations straight at n=2*p resulted higher by the Levinsona-Durbina method. Such method of prediction is named Prony Method. Instability at the prediction of future values of row is its failing. This method is not included therefore.

Other entrance data are:
- number of the last bar in past data
- amount of past bars of used for the prediction of future values
- order of linear model as fraction from the amount of past bars (0..1)
- amount of future bars in the prediction
- maximal amount of frequencies for the Method 1 (0 chooses all frequencies)
- error of calculation of frequencies for the Method 1 (>0.001 can not meet)
- number of weighing function for the Method 2 (0=Rectangular 1=Hamming 2=Parabolic)

An indicator draws two lines: a siniya line displays the costs of model on training bars, a red line displays the predicted future prices.

Examples

Method 1 (extrapolation of the Fure row)

Method 3 (the Burg method)

Method 6 (Modified Covariance Method)