Race 8 · Chantillypool vs model · Sat 10 Oct 2026
pool sharemodeledge, pts
  1. 3NOCE DE RUBISWON-7.518% · 26%
  2. 11UTXA (IRE)-3.718% · 21%
  3. 8NONNABELLA-2.39% · 11%
  4. 15ROMANCE MARINE-1.08% · 9%
  5. 12RUBENESQUE-2.67% · 9%
  6. 10MARQUES+0.76% · 5%
  7. 5SIMANCAS-0.95% · 6%
Won by NOCE DE RUBIS, our #1 — the pick landed · record

French racing picks — Sun 11 Oct 2026

Free horse racing picks for today's French cards: tote pool share plus sixteen form layers measured on 1,312 races, for every French track carried on the Turkish tote. Our ranks are written before the first race and scored after it: the record.

7 days on file · walk-forward test on 712 races out of sample · how it scored

Sun 11 Oct 2026No card published yet · the next one appears the evening before racingWritten Sun 11 Oct 2026, 09:08 Paris time · permalink

Unproven model — read this first

In a walk-forward test on 712 races the combined model did not beat the tote pool (-0.3 points on top-3 coverage). The ranks below are published as an open experiment: each one is written before the race and scored on the record, with the pool favourite as the benchmark to beat.

No French card up yet

TJK publishes foreign cards the evening before. Until then, browse the archive.

What the numbers mean

Pool share and edge

The Turkish tote (TJK) takes bets on French racing; each runner's share of that win pool is the market's opinion, published before the off. The model starts from that share and adjusts it with sixteen form layers (trainer, jockey, sire, damsire, owner, last finish, form, days since, head-to-head, track record, draw, rating, weight, age, depth of history). Edge is the model probability minus the pool-implied probability, in points. A positive edge means the model likes the horse more than the crowd does.

How the model was tested

Weights were fitted by conditional logit on 1,312 French races and tested walk-forward: each block scored with weights fitted only on earlier races, 712 races out of sample. On its own the model does not beat the pool (-7.2 pts top-3); combined with it, top-3 coverage moves -0.3 pts and likelihood improves. Verdict: failed. The gain is calibration, not coverage: the record page shows exactly how it performs.

Layer weights: piyasa +0.63 · form3 -0.37 · jokey +0.22 · sahip +0.22 · adres -0.17 · son_bitis +0.16 · antrenor +0.14 · draw -0.13 · n_gecmis +0.08 · hp +0.06 · aygir +0.05 · kgs +0.04 · siklet +0.03 · h2h +0.02 · damsire -0.01 · yas +0.00 · son_fark +0.00

Usable, thin, blind

A race is usable when at least 70% of runners have French form on file and the field averages two or more runs; thin at 40%; otherwise blind, where no verdict is given and the order shown is the market's. Before the pool opens (TJK opens it on race morning) every race is blind.

One discipline, five racing worlds:TJK Tahmin · TurkeySha Tin More · Hong KongUSA Pick Day · USASA Horse Race · South AfricaRace Franch · Francethe network →