Class AlphaValidation.WalkForwardResult

java.lang.Object
com.quantfinlib.alpha.AlphaValidation.WalkForwardResult
Enclosing class:
AlphaValidation
public static final class AlphaValidation.WalkForwardResult extends Object
All folds plus the aggregate in-sample vs out-of-sample comparison.

  • Constructor Details

    • WalkForwardResult

      public WalkForwardResult(List<AlphaValidation.Fold> folds, double meanInSampleIc, double meanOutOfSampleIc, double[] outOfSampleIcSeries, double outOfSampleTStat, double meanSelectionBias)
      Creates the value; array and collection arguments are copied.
      Parameters:
      meanInSampleIc - observation-weighted mean training IC of the winners
      meanOutOfSampleIc - observation-weighted mean test IC (equal to the mean of outOfSampleIcSeries)
      outOfSampleIcSeries - every test-window IC of the chosen candidates, in order
      outOfSampleTStat - Newey-West t-stat of that series (automatic lag)
      meanSelectionBias - observation-weighted mean of the folds' selection bias

  • Method Details

    • folds

      public List<AlphaValidation.Fold> folds()
      folds.

    • meanInSampleIc

      public double meanInSampleIc()
      meanInSampleIc (observation-weighted mean training IC of the winners).

    • meanOutOfSampleIc

      public double meanOutOfSampleIc()
      meanOutOfSampleIc.

    • outOfSampleIcSeries

      public double outOfSampleIcSeries(int i)
      Element i of outOfSampleIcSeries (every test-window IC of the chosen candidates, in order).

    • outOfSampleIcSeriesLength

      public int outOfSampleIcSeriesLength()
      Length of outOfSampleIcSeries.

    • outOfSampleIcSeriesCopy

      public double[] outOfSampleIcSeriesCopy()
      A fresh copy of outOfSampleIcSeries.

    • outOfSampleTStat

      public double outOfSampleTStat()
      outOfSampleTStat (Newey-West t-stat of that series (automatic lag)).

    • meanSelectionBias

      public double meanSelectionBias()
      meanSelectionBias (observation-weighted mean of the folds' selection bias).

    • efficiency

      public double efficiency()
      OOS/IS efficiency, NaN when the in-sample IC is not positive. Below about 0.5 the selection is mostly fitting noise.

    • deflatedInSampleIc

      public double deflatedInSampleIc()
      In-sample IC net of the expected best-of-N luck.

    • equals

      public boolean equals(Object other)
      Overrides:
      equals in class Object

    • hashCode

      public int hashCode()
      Overrides:
      hashCode in class Object

    • toString

      public String toString()
      Overrides:
      toString in class Object