paperportfolio-theorymean-varianceconditioningjf

Ferson & Siegel (2001) — "The Efficient Use of Conditioning Information in Portfolios"

The paper that derives the optimal nonlinear signal-to-weight mapping under mean-variance preferences. Journal of Finance, 56(3), 19–38.

Citation

Ferson, W.E. & Siegel, A.F. (2001). "The Efficient Use of Conditioning Information in Portfolios." Journal of Finance, 56(3), 19–38.

Core contribution

For a mean-variance investor who has access to a public signal s_t predicting future risk-adjusted returns, the unconditional-minimum-variance weight on the risky asset is a nonlinear function of the signal — specifically, an "S-curve" that is approximately linear in the middle and saturates (rolls back) at the extremes. This is the theoretical basis for the Ferson-Siegel Optimal Weighting used in Moskowitz Sabbatucci Tamoni Uhl 2025 Nonlinear TSMOM.

Intuition

  • The unconditional variance penalty of placing weight w scales as w^2 \sigma_\varepsilon^2 (where \sigma_\varepsilon^2 is the signal-noise variance).
  • The unconditional expected return contribution scales linearly in w.
  • For very large |s|, the quadratic penalty dominates the linear benefit, so optimal w saturates.

Why it matters

  • Provides the theoretical "why" behind Nonlinear Signal Mapping in any signal-driven strategy.
  • Validated empirically by the 2025 Moskowitz et al. paper, which shows a neural network trained to maximize Sharpe recovers a function nearly identical to FS.
  • A foundational reference in any modern CTA Strategy / Time-Series Momentum literature.

See also