How Well Did Kalshi Predict the Midterms?
We crunched the numbers on Kalshi's prediction market, where traders bet on election outcomes, exchanging over $2.5 billion. This was nearly double the amount traded in the 2024 general election markets. Kalshi's markets accurately predicted the winners in 95% of congressional and gubernatorial midterm races, according to a CBS News analysis.
Experts attribute the high accuracy to the large participant base and the availability of quality information.
"These are the biggest, most liquid prediction election markets we've ever seen," said Eric Zitzewitz, an economics professor at Dartmouth College, who has studied political prediction markets since 1880. "It's no surprise they were well-calibrated."
Traders bet on yes-or-no questions regarding event occurrences, with Kalshi profiting from transaction fees. For example, betting on Abdul El-Sayed winning Michigan's Senate race at 67 cents and successfully predicting his victory would yield a 33 cent profit, reflecting a 2/3 chance of winning at that time.
Here's a breakdown of Kalshi's market performance:
- House Races: 95% of favored candidates won.
- Senate Races: 91% of favored candidates prevailed.
- Governor's Races: 97% of favored candidates won.
On average, traders gave the eventual winner a 91% chance of success. Many midterm races were lopsided, explaining the high odds of 90% or more for some winners on Kalshi.
The favored candidate lost in 23 races, all of which were won by Republican candidates. Notable upsets included the South Carolina Senate race, where Kalshi gave Annie Andrews a 23% chance of winning, and the Texas governor race, where Gina Hinojosa beat incumbent Greg Abbott, priced at 33% on Kalshi.
To assess Kalshi's market calibration, we can analyze the performance of candidates based on their assigned odds. In a well-calibrated market, candidates with 60% odds should win 60% of the time, but Kalshi's results showed a slightly higher win rate for the "safe" races. Similarly, candidates in the "likely" range won more often than predicted.
While Kalshi's markets appeared accurate, it's important to note that race outcomes are not independent, as voter turnout and other factors can influence them.
"If lots of people turn out [to vote] for the Democrats, then all of the Democrats are more likely to win," said Tom Rietz, a finance professor at the University of Iowa, before the election.