The question was simple and old: can a sensible machine-learning model, built from public form data, beat the starting price? So we built one properly — two algorithms, XGBoost and LightGBM, reading around thirty pre-race signals per runner: the horse's own form, the going, the class, the field, the connections' recent record.
The first result looked brilliant, and it wasn't. Our spring 2026 backtest showed +15.73% on 330 bets, and for a day it felt like the answer. It was an artefact of selection bias: we had swept about 150 filter ideas and reported the shiniest one. On a window the sweep had never touched, the same filter returned roughly −17%. We published that failure rather than the mirage, and it is still the most useful thing this page has taught us.
So we rebuilt it the hard way. Walk-forward testing only (the model never sees a race after the one it is predicting), more data, more features, and isotonic calibration so its probabilities mean what they say. Then the exam that counts: one pre-registered rule, declared in advance, on an untouched two-month holdout — it returned −16.81% on 119 bets. Honest, negative, and exactly the point.
What that established is the foundation everything here stands on: the market is a formidably efficient forecaster, and a well-built model can read races almost as sharply and still lose at the price. Everything below is that finding, kept honest in public — every pick logged before the off, every result scored, updated daily. The full build story · How we test
Two tests, one honest question
This page is the scientific ledger: every pick the model makes, in every race, stored and scored for how sharp it really is. For the live daily test, its best picks put up against the bookmaker in real time, see Can it beat the bookies? →
One flat £10 on the model's own top pick in every British race it has priced, settled to industry starting price with the fallers counted as losses. No system, no staking trick, just the model's single best horse, day after day. This is how that balance would have felt in a real wallet.
On 550 race-days the balance finished up on 197 and down on 351. The best single day made +£242; the worst lost £258. Even its longest winning run (8 days) is shorter than its longest losing one (10 days). That is exactly what a market you can't beat looks like.
The running £ balance, every race-day since the record began. Green sections are days it rose, red days it fell. The gold line marks where backtesting ends and live, published-before-the-off predictions begin.
The same record split into calendar months. Green months won, red months lost. A good month happens; a run of them does not.
All-time against the most recent 90 and 30 days, including how the model's probabilities score against the market's on the same races (Brier score, lower is sharper). The market stays ahead.
| Window | Bets | Strike | ROI to SP | Profit / loss | Model Brier |
|---|---|---|---|---|---|
| All time | 7,530 | 25.5% | -11.8% | -£8,880 | 0.1025 |
| Last 90 days | 1,422 | 23.2% | -12.8% | -£1,816 | 0.1023 |
| Last 30 days | 936 | 20.7% | -15.5% | -£1,454 | 0.1002 |
Model Brier 0.1025 versus the market's 0.0932 on the same races: the model is well calibrated, just not as sharp as the price. That gap is the whole story, and the reason we show you this instead of selling you tips.
The same ledger, sliced the way the Tipster League slices its board. Pick a timeframe and the graph and numbers re-cut to it. £10 flat on the model's top pick per race, settled at SP.
Profit
−£1,102
ROI
-12.7%
Strike rate
21%
Bets
869
Staked
£8,690
Returned
£7,588
A running ledger of the model's latest top pick in each race: the selection, the market's price, and how it settled. Win or lose, every one is here, so you can see exactly how the model reads a race.
British Stallions Studs EBF Nursery
Five runs and two wins including last time out, with a perfect record at this course. The model estimates 38% while the market prices it at 42%, suggesting the
Best Ticket Deals Online At thirskraces Book Now Restricted Maiden Fillies' Stakes
Four runs on the card with a recent second, rated 69. The model's 45 per cent estimate sits notably above the market's 38 per cent assessment.
togetheragainfestival.co.uk Maiden Fillies' Stakes
Six runs in, but well beaten last time out, leaving little encouragement from recent form. The model estimates 36 per cent, well below the market's 56 per cent,
Wellman Cars Hamilton Handicap
A thoroughly seasoned runner with 38 starts and a recent victory, winning both at this track and twice from 13 at the trip. The model's 16% probability sits wel
By Chen Presidential Apprentice Handicap
Ten runs with a recent second place suggest a horse in reasonable form, rated 71. The model's estimate of 27 per cent sits noticeably below the market's 33 per
LSL Racing Horse Sales Syndication Leasing Apprentice Handicap
Twelve runs on the card with a recent second place, modest form overall from a horse rated 54. The model's estimate of 26 per cent sits just below the market's
Alex McWilliams Memorial Celebrations Restricted Novice Stakes
Four runs on the card with a recent second at a rating of 86. The model's 22% estimate sits well below the market's 59%, suggesting the market is pricing in con
Sell Your Business With Anderson Barrowcliff Handicap
Thirteen runs with a win at this trip and fourth last time suggest a capable handicapper rated 70. The model's 14% estimate sits above the market's 11%, indicat
EBF / UB40 ft. Ali Campbell 24th July Restricted Novice Stakes
Just one run to its name, a seventh, so the model has little to work with. At 33/1 the market is far more sceptical than the model's 13% estimate suggests.
Wellman Cars Private Hire Almada Mile Handicap
Don't Miss Summer Family Days At thirskraces Classified Stakes
Twenty-nine runs on the record with two wins at this trip, most recently fourth. The model's estimate of 9% sits above the market's 7%, suggesting the form comm
Fairplay Daily Price Boosts Handicap
Four runs without a win, though it has won two of six attempts at this trip. The model's estimate of 21% sits well clear above the 12/1 market price of 6%.
This is research, not tips. The model is a calibrated read on a race: it picks winners at a consistent rate and still loses to the starting price, which is the whole honest story. 18+, please bet responsibly.