fof-performance-attribution

Brinson attribution decomposes excess return into asset allocation, fund selection, and interaction effects using portfolio and benchmark weights and returns. Run the brinson command with holdings and benchmark files, and use Carino linking for multi-period results.

Use the Treynor-Mazuy model, which adds a squared market excess return term, or the Henriksson-Merton model, which adds a bull-market interaction term. A significantly positive coefficient on these terms indicates timing ability.

Brinson attribution requires holdings data and decomposes returns by allocation versus selection decisions. Factor regression needs only NAV series and estimates alpha and factor exposures via CAPM, Fama-French, or Carhart models. Cross-validating both gives more credible conclusions.

At minimum you need the FOF's NAV series for regression and risk metrics. Holdings data enables Brinson attribution, while factor return series (market, SMB, HML, MOM) and a risk-free rate series are needed for multi-factor regression and timing tests.

Alpha is a residual term that absorbs unmodeled style exposures and luck, so it cannot be equated directly with manager skill. Regression shows correlation, not causation, and an incorrect benchmark invalidates all attribution conclusions.