Steps
- Open the Allocation card on a method that reads one — see which methods use the covariance matrix.
- Press Change beside Covariance matrix (or Correlation matrix) and pick an estimate; the covariance list marks Ledoit-Wolf · constant correlation Recommended. What each one does is on covariance matrix.
- For EWMA, set the Decay factor; for Manual shrinkage, enter the Shrinkage intensity, which is required.
- Check the in-sample period is long enough for the estimator you chose — the card warns when it is not.
- On a split strategy, repeat for the other profile: Risk-On and Risk-Off keep their own estimate.
- On Min correlation, a chip under the method names any choice other than Sample: "Correlation: Pearson", "Correlation: Spearman", "Correlation: Kendall", "Correlation: EWMA" or "Correlation: Marchenko-Pastur".
Done. The method reads its volatilities and correlations through the estimate you chose, from the next backtest on.
Why does an estimator have no effect with a short history?
Two advanced estimators need enough history, and quietly fall back to Sample without it. Regime-conditional (HMM) needs about 126 trading days — an in-sample period of at least 6 months — and also falls back when calm and turbulent periods cannot be told apart. Nonlinear Ledoit-Wolf needs about 30 trading days, so at least 2 months. When the window is too short the app warns: "With this history the estimate has no effect", with an action to lengthen the in-sample period.
Why does a flat-priced instrument stop the backtest?
An instrument whose price stayed flat for the whole window has no measured risk, and most estimates leave it that way. With Sample, Ledoit-Wolf · constant correlation, EWMA, Manual shrinkage, Marchenko-Pastur or Regime-conditional (HMM), such an instrument stops the backtest when the method is Hierarchical risk parity, Hierarchical equal risk contribution, Nested clustered optimization or Max diversification. Ledoit-Wolf · identity and Nonlinear Ledoit-Wolf give it a small positive risk instead, so the backtest runs — Nonlinear Ledoit-Wolf only once the window holds about 30 trading days, below which it behaves as Sample.