Prediction markets doubt Le Pen's run-off more than her first round: polls and prices in the 2027 French presidential campaign

On 31 October 2026 the polls of the 2027 French presidential campaign, read through a Bayesian state-space model and carried to election day with an error calibrated on 2017 and 2022, gave Marine Le Pen 0.96 to reach the run-off and 0.89 to win it. A model of the Kalshi and Polymarket prices gave 0.79 and 0.42. A poll assumes a line-up, while a price includes the chance that a candidate stands, which a court case and the sponsorship rule left open. In an exploratory analysis we split the log ratio of the market and poll-implied probabilities of winning exactly into ballot access, the first round given a place on the ballot and the run-off given qualification. We also weight the poll-implied probability over line-ups drawn from ballot-access prices. At the cut-off the run-off term is 75 per cent of the gap on Le Pen (90% interval 70 to 83), ballot access 17 and the first round 8. The run-off is the largest term on either venue's raw prices and, in every month from July, under every variant. Under the weighting her qualification premium is close to zero. For second place, the markets rate only Raphaël Glucksmann below the polls under every price reading we ran, a pattern we treat as a conjecture. Prices moved on the day of the appeal ruling. Weekly tests, able to detect only large leads, found a lead for 1 of the 11 candidates. A registered evaluation follows the run-off of 2 May 2027.