What Is the Information Coefficient (IC)?
Let me explain the information coefficient (IC) directly to you—it's a measure I use to evaluate the skill of an investment analyst or an active portfolio manager. The IC shows how closely the analyst's financial forecasts match actual financial results. It ranges from 1.0 to -1.0, where -1 means the forecasts have no relation to actual results, and 1 means they matched perfectly.
Key Takeaways
You should know that the IC evaluates the skill of an investment analyst or active portfolio manager. An IC of +1.0 indicates a perfect prediction of actual returns, while 0.0 shows no linear relationship, and -1.0 means the analyst always fails at correct predictions. Remember, the IC is not the same as the Information Ratio (IR), which measures a manager's excess returns against the risk taken.
The Formula for the IC
Here's the formula you need: IC = (2 × Proportion Correct) - 1, where Proportion Correct is the proportion of predictions made correctly by the analyst.
Explaining the Information Coefficient
The IC describes the correlation between predicted and actual stock returns, and it's sometimes used to measure a financial analyst's contribution. An IC of +1.0 indicates a perfect linear relationship between predicted and actual returns, 0.0 indicates no linear relationship, and -1.0 shows the analyst always fails at correct predictions.
If an IC score is near +1.0, it means the analyst has great skill in forecasting. But in reality, if 'correct' means matching the direction (up or down) of actual results, the odds are 50/50, so even an unskilled analyst could have an IC around 0, with half right and half wrong. A score close to 0 means the analyst's skills are no better than chance, and scores near -1 are rare.
Again, don't confuse IC with the Information Ratio (IR), which measures a manager's skill by comparing excess returns to risk taken. Both IC and IR are components of the Fundamental Law of Active Management, which states that a manager's performance (IR) depends on skill level (IC) and how often it's used (breadth).
Example of the Information Coefficient
Take this hypothetical example: if an analyst made two predictions and got both right, the IC is (2 × 1.0) - 1 = +1.0.
If the predictions were right only half the time, then (2 × 0.5) - 1 = 0.0.
If none were right, it's (2 × 0.0) - 1 = -1.0.
Limitations of the Information Coefficient
The IC is only meaningful if the analyst makes a large number of predictions, because with few predictions, random chance can explain the results. For instance, two correct predictions give +1.0, but if it stays near +1.0 after dozens, it's more due to skill than chance.
