What Is Purged cross-validation?

Purged cross-validation is a time-series validation method that removes training observations whose information sets or label windows overlap a test fold. It is particularly useful when labels are based on forward returns or events spanning several bars, because otherwise the training and test sets can share information about the same future market movement.

Purging is applied separately within each fold and is often used with an embargo, a buffer of observations immediately following the test interval that is also excluded from training. It addresses one form of leakage but does not replace point-in-time data, a properly held-out test period, or realistic backtesting assumptions.

See also

• Embargo

• Cross validation

• Out-of-sample testing

Research and literature

• Marcos López de Prado, Advances in Financial Machine Learning, Wiley, 2018.