An autoregressive (AR) (or linear prediction) model is given by:
x[n] = -Sum(a[i]*x[n - i], i = 1..p)
where:
- x[n] is the predicted value of a time series;
- x[n-p]..x[n-1] are known past values of the same series;
- a[1]..a[p] are the model coefficients, and p is the model order.
The model coefficients a[1]..a[p] can be fitted to the past data by a variety of methods. This indicator uses the Burg method.
The inputs of the indicator are:
- UseDiff - a boolean switch to use price differences instead of prices themselves
- Ncoef - number of model coefficients (model order)
- Nfut - number of future bars
- kPast - number of past bars in increments of Ncoef (must be >=1)
The indicator plots two curves: the blue curve represents the model outputs during its fitting, the red curve shows predicted future prices.
UseDiff=false:
UseDiff=true:
