Signal Combination and the Information Coefficient

Once multiple alpha signals have been constructed and neutralised, the key question is how to combine them. Averaging signals ignores their relative quality; using the strongest signal ignores diversification benefits. The information coefficient (IC) — the rank correlation between a signal and the subsequent realised return — provides an objective measure of each signal's predictive value. IC-weighted linear combination is the industry standard for blending signals, and Grinold's Fundamental Law of Active Management (FLAM) shows why combining many weak signals can produce a strong composite: the Information Ratio scales with the square root of the number of independent bets.

• define the IC and explain what it measures;
• compute the IC for a cross-sectional signal using rank correlation;
• derive the FLAM formula $\text{IR} \approx \text{IC}\sqrt{N}$ and interpret its assumptions;
• implement IC-weighted linear signal combination;
• diagnose signals with unstable or negative IC and decide when to drop them.

1. 1. The Information Coefficient
2. 1. The Fundamental Law of Active Management
3. 1. IC-weighted signal combination
4. 1. IC decay and signal selection