The information coefficient is the correlation between a signal $g(t)$ and returns $r(t)$. I’m hoping to build some practical intuition on the information coefficient.

Similar to the notion of effect size in statistics, can we define an effect size for the information coefficient?

One way to answer this is by specifying ranges and labels, such as “0.00 to 0.03 - weak”. A brief intuitive description of the label could also be helpful to make sure we are on the same page, such as “weak: usually not worth further investigation”.

A more precise answer might map ranges to an estimated probability that the signal will be profitable, such as “0.00 to 0.03 - 25%”.

Obviously this question doesn’t have one straightforward answer, but I’m hoping for a reasonable guesstimate answer based on their intuition and market experience.

  • $\begingroup$ bump - I would like to bump this post $\endgroup$
    – Lmnop
    Sep 10 at 11:35


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.