Information Coefficient

The Information Coefficient (IC), introduced in the Grinold-Kahn active-management framework, is the rank or Pearson correlation between a manager's forecast and the realised return. For a systematic strategy, IC is the single number that quantifies "does the signal predict anything?".

Formula

IC = corr( forecast, realised return )

Computed per period across the cross-section of assets. Rank correlation (Spearman) is preferred to Pearson when outlier returns dominate.

Grinold's Fundamental Law

IR ≈ IC × √Breadth

Information Ratio scales linearly with IC and with the square root of Breadth (number of independent bets per year). Practical implication: a manager with modest forecasting skill (IC = 0.05) can still deliver competitive IR by making many independent bets.

Reading the number

• ·IC = 0 — no forecasting skill. The signal is uncorrelated with returns.
• ·IC = 0.02-0.05 — typical for the median systematic equity factor.
• ·IC = 0.05-0.10 — strong factor, e.g. value or momentum on the full cross-section.
• ·IC > 0.10 sustained over years — extremely rare; usually points to lookahead bias or in-sample fitting.

How MEDGE Capital uses IC

IC is not directly exposed in the MEDGE UI today because the user provides weights, not forecasts. For systematic users running tilt strategies on the Compare module, the implied IC can be inferred from the realised Information Ratio and the Breadth of the strategy — a useful diagnostic for whether outperformance is signal or luck.