Image Credit: “Fortune Teller, State Fair, Donaldson, Louisiana,” 1938, Russell Lee, Library of Congress
By Jason Zweig | Sept. 25, 2015 6:23 p.m. ET
After the humiliating failure of the U.S. intelligence community to predict correctly whether Saddam Hussein’s Iraq had developed weapons of mass destruction, the federal government sponsored a tournament to try developing better forecasting techniques. The success of one of the teams, co-directed by Wharton School researcher Philip Tetlock, can teach investors useful lessons about how to think about the financial future. Here are some of those principles, adapted from his new book “Superforecasting”:
1. Establish a Base Rate.
Compare. Establish the “base rate,” the typical historical frequency or severity of the outcome you are trying to forecast, by picking a reference class of similar past events. Very few situations are unique; think deeply enough about the instance at hand, and you will likely discover a set of historical precedents for it. Instead of guessing how stocks will do if interest rates rise by one percentage point, for instance, look at the full sample of all such past interest-rate increases, take the average of subsequent stock performance and use that as your base-rate assumption.
2. Be specific.
Try to flush your ignorance out into the open: Don’t just say, “I think emerging-markets stocks are cheap right now.” Instead, say “There is an 80% probability that emerging-markets stocks will go up at least 25% in the next 12 months.” When significant new information comes in, adjust your forecast incrementally.
3. Consider the Opposite.
Invert. Consider the opposite and ask what information you would need if you were trying to make a case that emerging-markets stocks will go down at least 25% in the next 12 months. Then seek it out, absorb it and test whether you need to revise your original forecast.
4. Cast a Wide Net.
Find a variety of sources for news and analysis, varying widely in their viewpoints, vested interests and geographic origin. Discount the sources that use the self-reinforcing language of overconfidence: Words like “furthermore,” “clearly” and “certain” or “impossible.” Heed most closely those sources that admit uncertainty with words like “however,” “but,” and “on the other hand.”
5. Measure Everything.
Record in detail the reasoning behind your forecasts. When you turn out to be right, don’t just celebrate; check your records to see whether you were right for the reasons you predicted or whether you just got lucky. When you turn out to be wrong, compare this error to the other times you were mistaken, looking for patterns to correct. If you don’t measure every meaningful aspect of your forecasts, you will never improve.