Better Forecasting Starts With Admitting You�re Wrong

Better Forecasting Starts With Admitting You’re Wrong

Related Symbols
TSPA

Good forecasting starts with a willingness to admit mistakes, according to Dan Gardner, co-author of “Superforecasting: The Art and Science of Prediction.”Key Takeaways: Superforecasters made frequent but small updates to their predictions, avoiding both stubbornness and overreaction to new information. Starting with category averages before examining specific details often produces better forecasts than focusing on particulars alone. AI models already beat the general public on forecasting questions and may rival top human forecasters by 2027 or 2028. Gardner made that case on The Angle, a podcast from T. Rowe Price Group, Inc (TROW). Intellectual humility is “absolutely foundational to good forecasting judgment,” he told host Justin Thomson, head of the T. Rowe Price Investment Institute. His book grew out of a forecasting tournament that U.S. intelligence officials had created after no weapons of mass destruction turned up in Iraq. Five university research teams competed to predict real-world geopolitical events, Gardner explained. Co-author Philip Tetlock’s team, the Good Judgment Project, recruited volunteers to make those forecasts. According to Gardner, only 1% to 2% consistently ranked at the top. Those habits can be taught, though, which makes them useful to anyone managing money. Volunteers who read one sheet of basic tips just once improved their forecasting accuracy by 10% over a full season, Gardner said. See more: How TCAF’s Research-Based Active ETF Approach Stands Out The first tip is to start with the base rate, or how common an outcome is across a whole category. Someone estimating the cost of a new bridge should check the average cost of bridges before studying their own project, Gardner explained. Some active stock funds work in a similar order, starting with a broad index before adding company-level judgment. The T. Rowe Price U.S. Equity Research ETF (TSPA B), for example, uses the S&P 500 as its baseline. About 30 research analysts then decide which stocks to overweight or underweight, according to T. Rowe Price. Sector and industry weights stay close to the index, with limits on how large any single position can get. TSPA launched in June 2021 and holds $4.7 billion in net assets as of October 20, according to the firm. Its expense ratio is 0.34%.Small Forecasting Updates Beat Big SwingsKnowing when to change a forecast, and by how much, is a challenge Thomson said he sees in investing as well. Portfolio managers tend to fall short either by reacting to every data point or by stubbornly sticking to a view, he added. Superforecasters updated far more often than other participants, but in small steps, Gardner said. That kept the value of older information while adding new data. Most people instead resist new evidence until it piles up, then throw out their view entirely. Gardner pointed to economist John Maynard Keynes, once called the “consistently inconsistent Mr. Keynes” in a Life magazine profile. The jab became a point of pride for Keynes, who Gardner said was also “a fantastic investor.” Artificial intelligence is now closing in on those skills. Large language models trailed the general public on forecasting questions in 2023 but have since passed it, Gardner noted. If that pace holds, they could match superforecasters around 2027 or 2028. That outlook has Gardner betting on a different skill. “The value of answers is going to decline,” he said. “The value of high-quality questions is going to soar.” For more news, information, and analysis, visit our Active ETF Content Hub.

Better Forecasting Starts With Admitting You’re Wrong

More Commentary

Search Commentary