Horse Racing AI — Common Questions
Honest answers to the questions punters actually ask about AI-driven horse racing intelligence. If you want to see how our signals perform, visit the Track Audit or the Accuracy page.
What is Capall?
Capall is the world's first data-backtested AI horse racing intelligence site for UK and Irish racing. It publishes calibrated ratings and value flags for every runner on every card, with the methodology, the backtest evidence and a live performance ledger all public.
How does the AI rate horses?
Two ratings per runner. CR is a pre-race rating set the night before, scored 0–100. CMR is the race-day rating, also 0–100, that re-fits as the live market moves into the off. Both come from the same calibrated AI engine. Full feature list and validation method live on the methodology page.
Is the AI's track record public?
Yes. Every settled signal is on the live performance ledger. Losing months are listed exactly the same way as winning months — no cherry-picking, no hidden picks. The 12-month walk-forward backtest is on the backtest page.
What's the difference between CR and CMR?
CR is the pre-race form picture — what the engine thinks of the horse on the night before, before any race-day market information. CMR is the race-day picture — CR re-weighted by what the live market is doing into Live Lock at T-2 minutes. CMR is what most users follow on the day.
What is a 'value flag'?
A value flag is the engine saying it thinks the market has the horse priced wrong in a way that gives a positive expected return. It's not a tip — it's a flag for inspection. The bucket label (HERO, VALUE, UNICORN) tells you which tier of value the engine believes it has spotted.
What is Live Lock?
Live Lock is the moment the engine stops re-rating a race. It fires at T-2 minutes (two minutes before the off). After lock, the CMR rating, the verdict and the bucket are frozen. Everything published on the live performance ledger is what was on screen at the lock moment, not what the engine thought afterwards.
Does the AI cover every UK and Irish race?
Yes. Every runner on every flat and jumps card across the UK and Ireland gets scored.
Is Capall free?
There's a free tier with delayed signals (verdicts shown 48 hours after lock) so anyone can audit the record. Real-time signals, the full race card view and Ask Cory live behind the paid Capall plan.
What is the 'Honesty Layer'?
The Honesty Layer is the set of rules that force the engine to publish PASS — to go silent — when the conditions for a clean signal aren't there. Drift past a certain band, BSP misalignment, calibration drift, an old model — any of these can force a PASS even on a horse the engine likes. It's the thing that stops the site from being noise dressed up as confidence.
Is this betting tips?
No. Capall publishes ratings and value flags — the engine's view of where the price is wrong. It does not stake-size, it does not tell you to bet, and the methodology is published so you can reconstruct every number. Whether to back, lay or pass is your decision.
Who built Capall?
Capall was built with Cory Webber, Julian's son-in-law. The full story lives in the methodology page footer and on the about page. The product is run as a small focused team that ships only what passes the engine's own honesty rules.
What is BSP?
BSP is the Betfair Starting Price — the lock-in price on the Betfair exchange when the race goes off. It's the most honest single price reference for a horse because it's the price actual money was matched at. Both the backtest and the live performance ledger settle outcomes at SP and at BSP side-by-side.
How often is the engine refit?
On a rolling schedule, with calibration tracked between refits. If calibration drifts past the allowed band the engine is taken offline until it's been re-fit and reviewed. Refit dates are noted in the methodology decision ledger.
Where does the data come from?
The official Racing API feed for race cards and form, the Betfair exchange for live market data and BSP, plus our own race-by-race tracking layer. The full source list and feature engineering steps are on the methodology page.
What does 'data-backtested' actually mean?
It means every change to the engine has to clear walk-forward validation against historical races before it goes live — fitted on a rolling window, scored on the next slice, settled, then advanced. No future data ever leaks back into the training of an earlier slice. The numbers behind that process are on the backtest page.
Is the site safe to use? What about responsible gambling?
Capall is intended for adults 18 and over. Bet only with money you can afford to lose. If gambling is becoming a problem, please visit BeGambleAware.org. The site does not encourage chasing losses and publishes losing months as openly as winning ones.
Can I get a refund? How do I cancel?
14-day refund window from first charge, no questions, email cory@capall.net. Cancel any time from your settings page — one click, two screens, no retention loop. Your access stays live until the end of the period you've already paid for.