Wall Street Looks to Superforecasting

The importance of accurate forecasting is leading some firms to using so-called superforecasting—a statistical approach in which “large numbers of estimates are averaged to find the most likely answer.” This according to a recent article in The Wall Street Journal.

“In finance,” the article asserts, “a good forecast can mean the difference between bumper earnings and bankruptcy.” Superforecasting involves a complex process whereby individual predictions are “combined and weighted, based on timeliness and past accuracy, until finally all of the results are melded into one forecast.” The methodology, the article explains, attempts to exploit the “wisdom of crowds.”

The idea began in 2011 at a forecasting tournament in which thousands of amateurs competed against seasoned forecasters (the amateurs won). Those amateurs who scored in the top 2% now make predictions for Good Judgement Inc. (a commercial spinoff of the original experiment), which has conducted workshops for financial firms such as BlackRock and Oaktree.

Another company that uses this approach is Estimize, a New York-based financial forecasting firm that collects estimates from as many as 70,000 contributors when forecasting quarterly earnings. Each estimate is then weighted using a confidence algorithm that “takes account of historical accuracy within a forecaster’s sectors, the freshness of the estimate, and the number of predictions he or she has made.”