Information Coefficient vs Correlation Coefficient

Hi everyone,

I am a bit confused by the definition of IC.

According to Wiki and Investopedia it is similar to Corr, but it is defined between 0 and 1.

According to the L2 CFAI curriculum it is identical to the Corr, and also defined between -1 and 1.

Can anyone clarifiy?



Informational coefficient from the portfolio management section is indeed -1 to 1, calculated by IC = 2(% correct) - 1. It serves as a proxy for how correct a manager is.

Thank you capitalism.

I am not sure how to interpret your formula : IC = 2(% correct) - 1. , care to elaborate?

If IC is identical to the correlation coefficient (Corr) then the formula should be cov/std_a*std_b, right?

And again, if it is indeed defined between -1 and 1, why are Wiki and Investopedia claiming that it is defined between 0 and 1?



Maybe you should read the text a little bit more, wikipedia and investopedia aren’t the CFAI that is defining these. So, why you would try to use them as “word is born” is beyond me. It is only related to correlation coefficient in the sense that it measures a relationship between predicted returns / actual returns.

Well typically the information in the CFAI curriculum and other sources are not contradictive, which is why this caugt my attention. I am simply trying to reconcile these different notions, or at least understand why Wiki and Investopedia are using a different definition.

If you use your formula: IC = 2(% correct) - 1

It appears to me that you obtain the same result as you would from computing the correlation between predictions and actual returns. Say a Portfolio Manager (PM) is correct 50% of the time, so randomly guessing, your formula yields IC=0. For 100% correct, you would get IC=1, and 100% incorrect, you would get IC=-1. And corresponding values in between.