🔬 Transparency
The four rules we hold ourselves to
- Published before the off. Picks go up before the first race of the day and we don't change them — and the daily hash chain lets you check that no earlier day was edited.
- Settled publicly. Every pick gets a winner/loser stamp after the race — wins and losses both, in the open.
- Losers stay visible. Bad days don't disappear. No cherry-picked highlight reel.
- No edits after the off. If we got it wrong, you can see it.
We’ll never sell you a profit promise — and here’s the honest reason
We back-tested every scoring variant against tens of thousands of real results. The honest finding: the AI ranks a race sensibly, but it tracks the market rather than beating it — its top picks don't turn a profit at the prices you can actually get, and a shortlist of the favourites does the job just as well. So we don't sell tips, a “verified P&L”, or a profit promise. Anyone who does is guessing or selling you something.
What this site is for is the analysis underneath every race — the data and tools to help you form your own view:
- 📊 Racecards — every runner with RaceIQ sectionals, pace & form
- 🧬 Breeding Lab — sire/damsire suitability for today's going & trip
- ⚡ Edges — build your own angles across 40+ data fields
- 🏟️ Track Bias — draw, going & pace bias by course
The full method and per-variant detail is on the methodology page. Gamble responsibly — BeGambleAware.
📊 AI Score reliability
Sort every UK & Irish runner this year into 10 equal bands by the AI's rating, then check how often each band actually won. A rising staircase means the score genuinely tells stronger runners from weaker ones.
🧪 Our models’ record
The models we actually put in front of you — the flagship Top Pick and the AI Score on every runner — tracked daily at £10 level stakes and published as-is, wins and losses. Tap 📈 for the day-by-day cumulative curve and monthly table.
⛓️ Proof of Picks
How the daily pick hash chain works (technical)
Every day's top picks are hash-chained before the first race — each entry's SHA-256 hash includes the previous day's hash, so no entry can be altered retroactively without invalidating every subsequent hash. That lets anyone recompute the chain and confirm no earlier day's picks were changed. It's an internal consistency check — it doesn't by itself prove the publication time.

