Information Coefficient - IC
Categories: Metrics
An information coefficient, or IC, is the difference between what was predicted and what actually happened. We should probably narrow that down a bit and specify that we’re talking about financial portfolio performance...and not the fact that we predicted we’d live in a castle in Italy by the time we were 30, but ended up in a condo in Toledo instead.
An IC can range from -1.0 to 1.0. The higher the number, the closer the predictions match real life. So if an investment analyst perfectly predicted a company’s stock behavior, the IC on that would be 1.0. If they correctly predicted how one stock would behave but totally missed the mark on another one, their overall IC would be 0, because the average of 1.0 (being totally right) and -1.0 (being totally wrong) is zero. In math terms, the equation looks like this:
IC = (2 x C) – 1
“C” is the proportion of predictions that turn out to be correct. So if our analyst makes 20 predictions and only one ends up being right, that would look like this:
IC = (2 x .05) – 1
Our analyst’s IC is -.9, which indicates that maybe they’re not so great at making predictions...and we might not want to rely too heavily on their financial advice.