Information Coefficient
The Information Coefficient (IC) measures the correlation between a manager's or model's forecasted returns and the actual realized returns, ranging from -1 to +1, and serves as the primary measure of forecast skill in quantitative investing.
The Information Coefficient is one of the most important concepts in quantitative portfolio management and is central to the Fundamental Law of Active Management, developed by Richard Grinold and Ronald Kahn. The Fundamental Law states that the information ratio — the reward per unit of active risk — is approximately equal to the IC multiplied by the square root of the breadth (the number of independent investment bets made per year).
An IC of 1.0 would represent perfect forecasting ability — every prediction is correct. An IC of 0.0 means the forecasts have zero predictive power relative to outcomes, essentially random. In practice, skilled systematic managers typically achieve ICs in the range of 0.03 to 0.10 — seemingly low, but applied across hundreds or thousands of positions over many periods, even an IC of 0.05 can generate meaningful and consistent alpha.
IC is typically measured using rank correlation (Spearman rank IC) rather than standard Pearson correlation, because return distributions are not normal and outliers can distort Pearson correlations. Rank IC compares the ranking of predicted returns to the ranking of realized returns across the investment universe. The IC Information Ratio (ICIR) — the mean IC divided by the standard deviation of IC over time — measures the consistency of the signal, which is as important as its average strength.
Signal decay is a related concept: the IC of a forecast tends to decline as the holding period extends beyond the horizon for which the signal was designed. A signal designed to predict five-day returns might have an IC of 0.07 over five days but near zero at 60 days, meaning the forecast has no value over longer horizons. Understanding signal decay guides portfolio rebalancing frequency.
For fundamental active managers who do not use explicit quantitative models, the IC concept still applies implicitly: the correlation between the analyst's conviction about a stock and its subsequent return. Research shows that even professional analysts, on average, have ICs only modestly above zero, with high variation across individuals, which is one reason systematic portfolio construction techniques are valuable even in qualitatively-driven investing.