Information Coefficient and Information Ratio

Hey everyone,

if the IC is defined as a correlation between predictions and actual returns, then it is defined between -1 and +1 correct? I somehow find contradiction definitions, that it is 0-1 or -1 to +1. Because this does matter when we plug it into the IR computation, i.e. if we are interested in high IRs then a negative IC would make the IR negative. But then again, I am wondering, is a negative necessarily a bad thing, since it would still allow us to draw conclusions from the predictions.

Thanks

Tartaglia

yes, IR can be negative.

IR is active return over active risk

active return can be negative as it’s excess return over the benchmark