Prix de l'Arc de Triomphe

A win percentage of 83.1% is not a verdict the model delivers lightly, and it is worth sitting with what that figure actually means before accepting it as settled. Across 1,000 simulated Arcs run on the balanced lens — which explicitly permits chaos — Daryz failed to win on only 169 occasions, and in every single one of those 169 reversals it was Kalpana who beat it. That is a remarkably binary distribution for a 16-runner field. The model is, in effect, telling us that this is a two-horse race in which one of the horses is a prohibitive favourite, and that the rest of the field — regardless of trainer reputation, jockey seniority, or international profile — is racing for experience rather than victory. The market, at 15/8 and 5/1, is saying something similar, but the multiverse's version is considerably starker than the bookmakers have priced. Where the model and the market diverge most sharply is at the top of the book. Kalpana's 5/1 reflects a genuine challenger who wins one race in six according to the layers — the multiverse returns 16.0%, which is broadly consistent with that. The interesting tension is with the horses priced between 25/1 and 80/1. Minnie Hauk at 25/1 under William Buick for Aidan O'Brien, Friendly Soul at 25/1 for the Gosden yard, Saddadd at 40/1 for Roger Varian — these are not unfancied names, and the market is allocating meaningful probability to their winning. The model, having run 1,000 iterations, gave each of them precisely zero wins and zero top-three finishes. That is a significant disagreement. Whether the model is identifying form superiority the market has underweighted, or whether the balanced lens is failing to capture the idiosyncratic conditions — ground, draw, pace collapse — that allow a 25/1 shot to steal an Arc, is the legitimate uncertainty here. Maltese Cross, at 0.9% and a mean finish of 3.14, occupies the one genuinely interesting position in the distribution. The model sees a horse capable of troubling both principals on a small but non-negligible number of occasions — the 76.0% top-three rate is a meaningful figure for a runner who never wins. If there is a live each-way argument anywhere in this field, the model would locate it there, and almost nowhere else. The Japanese raiders, historically capable of landing European prizes, find no traction whatsoever across 1,000 iterations — a finding that should give pause to anyone constructing an exotic wager around Meisho Tabaru or Admire Terra.