Class SharpeValidation

java.lang.Object

com.quantfinlib.backtest.validation.SharpeValidation

public final class SharpeValidation
extends Object

Sharpe ratio significance tests (Bailey & López de Prado):

• Probabilistic Sharpe Ratio — the probability the true Sharpe exceeds a benchmark, adjusting for track length and non-normal returns (skew, kurtosis).
• Deflated Sharpe Ratio — PSR against the Sharpe you'd expect from the best of N random trials: the multiple-testing haircut for a strategy picked from a parameter grid.

All Sharpe inputs are per-period (not annualized) with observation count nObs.

Method Summary

Modifier and Type	Method	Description
static double	deflatedSharpe(double observedSharpe, double[] trialSharpes, int nObs, double skewness, double kurtosis)	Deflated Sharpe: PSR of the winner against the expected-max benchmark implied by all the parameter combinations that were tried.
static double	expectedMaxSharpe(int trials, double varianceOfTrialSharpes)	Expected maximum Sharpe among trials independent zero-skill strategies whose Sharpe estimates have the given cross-trial variance.
static double	minTrackRecordLength(double observedSharpe, double benchmarkSharpe, double skewness, double kurtosis, double confidence)	Minimum track record length (Bailey & López de Prado): how many periods of THIS performance are needed before probabilisticSharpe(double, double, int, double, double) would clear confidence that the true Sharpe exceeds the benchmark — the allocator's question ("how long until this manager's record means something?")
static double	probabilisticSharpe(double observedSharpe, double benchmarkSharpe, int nObs, double skewness, double kurtosis)	Probability the true Sharpe exceeds benchmarkSharpe, in [0,1].

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Details

probabilisticSharpe

public static double probabilisticSharpe(double observedSharpe,
 double benchmarkSharpe,
 int nObs,
 double skewness,
 double kurtosis)

Probability the true Sharpe exceeds benchmarkSharpe, in [0,1].

expectedMaxSharpe

public static double expectedMaxSharpe(int trials,
 double varianceOfTrialSharpes)

Expected maximum Sharpe among trials independent zero-skill strategies whose Sharpe estimates have the given cross-trial variance.

deflatedSharpe

public static double deflatedSharpe(double observedSharpe,
 double[] trialSharpes,
 int nObs,
 double skewness,
 double kurtosis)

Deflated Sharpe: PSR of the winner against the expected-max benchmark implied by all the parameter combinations that were tried. Values near 1 mean the edge survives the multiple-testing haircut; below ~0.95 the "discovery" is likely selection bias.

Parameters:

trialSharpes - per-period Sharpe of every trial in the search (including the winner)

minTrackRecordLength

public static double minTrackRecordLength(double observedSharpe,
 double benchmarkSharpe,
 double skewness,
 double kurtosis,
 double confidence)

Minimum track record length (Bailey & López de Prado): how many periods of THIS performance are needed before probabilisticSharpe(double, double, int, double, double) would clear confidence that the true Sharpe exceeds the benchmark — the allocator's question ("how long until this manager's record means something?") in closed form:

  n* = 1 + (1 − γ₃·SR + (γ₄−1)/4·SR²) · (z_conf / (SR − SR*))²

Returns POSITIVE_INFINITY when the observed Sharpe does not exceed the benchmark — no track record length proves an edge the record does not show. Sharpe inputs are PER-PERIOD (not annualized), matching probabilisticSharpe.