Prediction markets favored Bass as polls pointed to Raman in Los Angeles race

Prediction markets repeatedly favored Karen Bass in Los Angeles’s mayoral race even as runoff polls generally put Nithya Raman ahead, an analysis by USC Annenberg Media found. The markets ultimately got the result wrong: Bass conceded to the City Council member on Nov. 5, when Raman had 54.7% of the vote with more than 95% counted, a lead of nearly 10 percentage points.

From May onward, polls collected by The New York Times consistently showed Raman ahead in the runoff, though some were sponsored by one of the candidates. UC Berkeley’s Institute of Governmental Studies found Raman ahead in its May and September surveys, before recording a 40%-40% tie in October. A UCLA Luskin School of Public Affairs-reported poll also gave Raman an 11-point advantage, although one-third of likely voters remained undecided.

The prediction-market picture was markedly different. Data showed Bass ahead from April through September, usually by about 20 percentage points. Her Kalshi odds peaked at 80% on June 2, the primary day when she advanced to the runoff, while her highest Polymarket price was 70.5% on May 27. Her odds had risen by nearly 20 points earlier in May around a debate involving Bass, Raman and Spencer Pratt.

Market prices fluctuated on the days Berkeley released new polls. Raman’s price briefly rose when Berkeley released its May poll and climbed by more than 10 points on Sept. 23, the date of its September survey. But Bass regained the market lead on Oct. 28, when Berkeley reported the 40%-40% result. In that poll, expectations differed from stated voting preference: 40% of respondents said they expected Bass to win, against 32% who expected Raman to prevail.

Both platforms put Bass more than 30 points ahead immediately before the Nov. 3 election. An OddsChecker snapshot published on Election Day put the margin closer, with Bass priced at about 58 cents and Raman at 42 cents on both Kalshi and Polymarket. It also found that Kalshi’s favorite had switched three times in the preceding 10 days.

As our October coverage documented, Raman temporarily held narrow leads in both markets during the final weeks, underscoring how sharply the pricing moved before Bass recovered the edge by Election Day.

Mark DiCamillo said the apparent conflict reflected two different measures. Prediction markets capture the expectations of people placing bets, he said, while surveys ask voters about their intended choices. Market prices are set through participants buying contracts, whose values represent the market’s consensus implied probability.

Wei Xuan cautioned that the mechanism did not give prediction markets an inherent accuracy advantage. Participants have an incentive to make accurate bets, Xuan said, but the result can depend on who takes part. He also warned that supporters could spend money to push up a candidate’s price to generate a favorable headline rather than to profit from the market.

Trading in the race was substantial. Kalshi’s final volume was nearly $150 million, while Polymarket’s was just over $18 million. Kalshi alone had exceeded $109 million by Nov. 3.