Racing questions, answered with data instead of promises

Every completed week of the public ledger as its own page: the calls, the winners, the return at the logged prices, and what the SP favourite did in the same races. New every Monday.

Ombudsman beat Calandagan in the Champion Stakes, Almeraq won the Sprint at 11/10, Mount Atlas the Long Distance Cup at 33/1 and Fifth Column the Balmoral at 22/1. Favourites 2 from 7, the draw, OddsLens's six calls settled in public, and 242 Ascot races on favourites, the draw and the big handicaps.

Pierre Royal 33/1 from stall 9 in the Cambridgeshire with the 4/1 favourite fourth, favourites 8 from 21 across the meeting and 1 from 8 in the handicaps, the three days race by race, OddsLens's 21 calls settled in public, and the course's numbers from 243 Rowley Mile races.

Pellitory 12/1 from stall 16 in the Gold Cup, favourites 4 from 23 across the meeting, the three Cups race by race, OddsLens's 22 calls settled in public, and the course's numbers from 210 Ayr races.

10–13 September. The St Leger at 3:35 on Saturday, last year's results race by race, favourites, the draw and the big handicaps in 314 Doncaster races — and a tracked call in every race.

Yes — about as well as the market can, which is the catch. What machine learning genuinely can and can't do, evidenced by a public record.

Why professionals judge betting by whether prices beat the close, not by last month's profit — with our own awkward pair of numbers as the worked example.

The field-size terms table, the 4-runner trap where each-way pays win only, and why one non-runner can delete your place.

Deleted losers, "advised at" fiction, VIP screenshots — the tricks, and the five verifiable things a genuine record must have.

18,925 races of data: favourites won 34.4% — and lost 10.7p per £1 doing it. And there is no price band where it turns a profit.

Racing's "safest" bet wins nearly six times in ten — and still loses money. Why a high strike rate is not the same thing as profit.

The bigger the price, the worse the bet — all the way down to −49.9% at 33/1 and bigger. Everything punters believe about where value hides, inverted.

High draws won 22% of races against a 33% fair share. The stall-by-stall numbers before Glorious week — with Chester, York and Ascot as method checks.

28 July – 1 August. The feature races day by day, and a tracked model pick in every race — losses included.