As the 2024 U.S. presidential election looms, political analysts are scrambling to predict the outcome of the contest between Kamala Harris and Donald Trump. Among the voices offering an unconventional take is Thomas Miller, a data scientist at Northwestern University. Known for his unorthodox but highly accurate predictions in the 2020 election, Miller uses a unique approach to forecast what he sees as a potential landslide victory for Kamala Harris.
In 2020, Thomas Miller’s innovative methodology, which relied on betting markets rather than traditional polling, proved incredibly accurate. He nearly pinpointed Joe Biden’s electoral win and precisely forecasted the outcomes of two critical Georgia Senate races that ultimately handed control of the U.S. Senate to the Democrats. Thomas Miller is applying the same methods to analyze the upcoming Harris-Trump showdown, and his findings are surprising and compelling.
Polls vs. Prediction Markets
Miller’s approach differs dramatically from standard polling techniques. While most political forecasters rely on surveys of likely voters, Thomas Miller uses data from political betting markets, where individuals wager their own money on which candidate they believe will win. He argues that these betting markets provide a more accurate prediction of election outcomes because they reflect real-time public sentiment. As he puts it, “polls are a snapshot of the recent past,” whereas betting markets offer a forward-looking perspective.
By tracking the daily changes in betting odds on PredictIt, a popular U.S. political betting site, Miller’s model captures voter sentiment fluctuations more precisely than traditional polls. For example, in the days leading up to the September 10 debate between Harris and Trump, his model showed that Harris had surged to a lead of over 400 electoral votes—a sharp reversal from earlier in the race when Trump appeared dominant.
Harris’s Lead and the Path to a Landslide
According to Thomas Miller’s model, as of mid-September, Harris is favoured by 55% of bettors, with Trump trailing at 45%. If these numbers hold, Miller predicts a landslide victory for Harris, reminiscent of Lyndon Johnson’s overwhelming defeat of Barry Goldwater in 1964. “We’re talking about a blowout,” Miller notes, suggesting that Harris could win key swing states like Michigan, Pennsylvania, and North Carolina.
While many traditional pollsters still see the race as tight, Miller’s markets-based model suggests otherwise. If his predictions prove as accurate as they were in 2020, the 2024 election may result in a resounding victory for Harris.