Skip to main content
aifinhub
Backtesting & Validation Formula

Information Coefficient Formula

The information coefficient (IC) is the correlation between a forecast and the realized outcome it predicts. For a quant signal it is the correlation between predicted and actual returns across the cross-section of assets, or between a ranked signal and forward returns. An IC of 0 is no skill; even a small positive IC, applied across many independent bets, can build a strong information ratio.

By AI Fin Hub Research · AI Fin Hub Team
Best Next MovePlaygrounds

Forecast Scoring Sandbox

Paste a forecast stream (probability + outcome) and see Brier score with full decomposition, log loss, reliability diagram, and bootstrap confidence.

CalculatorOpen ->

On This Page

Formula

Copy the exact expression or work through it step by step below.

IC = corr(forecast, realized) Rank IC = corr(rank(forecast), rank(realized)) (Spearman form) Link to active management: IR = IC x sqrt(breadth)

Variables

forecast

Forecast signal

The model's predicted return, score, or rank for each asset at the start of the period.

realized

Realized return

The actual forward return over the holding period. IC measures how well the forecast lined up with what actually happened.

rank(.)

Rank transform

Converting forecasts and outcomes to ranks before correlating gives the rank IC (Spearman), which is robust to outliers and the common choice for equity cross-sectional signals.

breadth

Breadth

The number of independent bets per year. The fundamental law of active management links IC to the information ratio: skill per bet times the square root of the number of independent bets.

Step By Step

  1. 1

    Collect paired forecasts and realized outcomes across the cross-section for a period.

    Predicted scores and forward returns for 500 stocks this month.

  2. 2

    Compute the correlation between forecasts and realized outcomes (use ranks for rank IC).

    The cross-sectional rank correlation is 0.04.

  3. 3

    Average the per-period ICs over the sample and assess their stability.

    Mean monthly IC of 0.04 with a positive but noisy time series.

  4. 4

    Translate IC into expected information ratio using breadth via the fundamental law.

    With breadth 120 independent bets per year, IR = 0.04 x sqrt(120).

Worked Example

Equity cross-sectional signal evaluated for skill

Mean IC per period

0.04

Independent bets per year (breadth)

120

The information ratio implied by the fundamental law is IR = IC x sqrt(breadth) = 0.04 x sqrt(120) = 0.04 x 10.954 = 0.438.

An IC of just 0.04, which sounds negligible, implies an information ratio of about 0.44 when applied across 120 independent bets a year. This is the core insight of the fundamental law: a tiny per-bet edge becomes a respectable risk-adjusted return through breadth. The caveat is that breadth must be genuinely independent, which crowding and shared factor exposure erode in practice.

Common Variations

Rank IC (Spearman): correlates ranks instead of raw values, the robust default for cross-sectional equity signals.
IC decay: tracking how IC falls as the forecast horizon lengthens, to find the signal's optimal holding period.
Risk-adjusted IC: correlating the signal with risk-adjusted rather than raw forward returns to credit only true alpha.

Try These Tools

Run the numbers next

Sources & References

Related Content

Keep the topic connected

Planning estimates only — not financial, tax, or investment advice.