Researcher ORCID Identifier
Prediction Markets vs. Sportsbooks: A Comparison of NFL Forecast Accuracy
Graduation Year
2026
Date of Submission
4-2026
Document Type
Open Access Senior Thesis
Degree Name
Bachelor of Arts
Department
Economics
Reader 1
Florian Madison
Rights Information
Dylan M Cotti
Abstract
This paper compares the forecast accuracy of Kalshi, a prediction market, and DraftKings, a sportsbook, using data from 285 NFL games in the 2025-26 season. I collect win probabilities from both platforms at three time points in the week before each game and measure accuracy using the Brier score. I find that Kalshi produces slightly more accurate predictions than DraftKings, but the gap is small and marginally significant. More surprisingly, both platforms become less accurate as the game approaches rather than more accurate. Once I control for differences in how each platform distributes its quoted probabilities, the accuracy gap largely disappears, suggesting the difference is more about how each platform quotes odds than about one having genuinely better information.
Recommended Citation
Cotti, Dylan, "Prediction Markets vs. Sportsbooks: A Comparison of NFL Forecast Accuracy" (2026). CMC Senior Theses. 4214.
https://scholarship.claremont.edu/cmc_theses/4214
