The Informed Test: AI that sees the market
In the informed arm, each AI gets the same racecard plus the market's implied chance for every runner— the crowd's collective view of the race, with the bookmaker's margin stripped out (it never sees the raw Starting Price it will be settled at). The question this answers is a sharp one: does handing an AI the market's wisdom make it a better tipster, or does it just make it copy the favourite?
Only the five chatbots run here. Our own Stablebet Model and The Favourite compete in the blind arm only — the model is built to be independent of the price, and an informed favourite would just be backing itself.
Research, not tips. Settled honestly to industry SP, wins and losses alike. 18+ · please gamble responsibly.
The informed leaderboard
1,000 informed picks settled| # | AI | Lab | Bets | Win% | Staked | Returned | Profit | Backed the fav | Profit trend |
|---|---|---|---|---|---|---|---|---|---|
| 1 | DeepSeek | DeepSeek | 200 | 27% | £200 | £166 | −£33.78(-16.9%) | 95% | |
| 2 | ChatGPT | OpenAI | 200 | 26% | £200 | £164 | −£36.44(-18.2%) | 98% | |
| 3 | Gemini | 200 | 27% | £200 | £163 | −£37.19(-18.6%) | 98% | ||
| 4 | Claude | Anthropic | 200 | 26% | £200 | £158 | −£41.69(-20.8%) | 100% | |
| 5 | Grok | xAI | 200 | 26% | £200 | £158 | −£41.69(-20.8%) | 100% |
Running profit, race by race
Each line is one competitor's cumulative profit or loss in the informed arm at £1 a bet, settled to Starting Price. It builds live.
What the numbers say
The read on 7 days and 2,245 settled picks, worked out live from the board.
Every AI finds the winner in about 26.0%–27.0% of races, within a whisker of each other, and every one still loses. The market prices those winners too short. Naming the likely winner is not the same as being paid enough for it.
They back the market favourite 95%–100% of the time, and went against a unanimous field only 4–12 times each. Most of the time, asking an AI is close to just backing the favourite.
Not enough: none beats the bookmaker's margin, and nobody is in profit.
Return per £1 staked, settled to Starting Price. The Stablebet model and backing the favourite are shown as baselines. Research, not tips.
Run blind, these same AIs did better — by an average of £14.60 per £100 staked — because they read the race instead of leaning on the favourite the market points them at.
Questions
What does the informed test measure?
Whether showing an AI the market's implied chances makes it a better tipster. Each model sees the racecard and the crowd's probability for each runner (with the margin removed), picks one winner before the off, and we settle at Starting Price. Comparing it to the blind arm shows what the market actually adds.
Does seeing the odds make the AI back the favourite?
The 'Backed the fav' figure on the board above answers it live. In practice, showing an AI the market pulls it heavily onto the favourite — which is exactly why the informed and blind arms diverge, and why 'informed' can turn out to mean little more than 'backs the jolly'.
Why isn't the Stablebet Model or The Favourite here?
Both are blind-only by design. The Stablebet Model is deliberately built without the market price as an input, so showing it the odds would collapse the comparison. The Favourite already IS the market's view, so an informed version would simply be backing itself. They run in the blind arm as the baselines.
Is the informed AI profitable?
The board is derived live and reports wins and losses honestly. Across the sample so far, seeing the market has not turned any AI into a winner — copying more favourites wins more races but at prices too short to cover the losers. Nothing here is a tip.
Gamble responsibly.This page is research and entertainment, not betting advice. No AI here beats the bookmaker's margin, and nothing on it is a signal to stake. Betting should never be a way to make money. If it is affecting you or someone you know, free and confidential support is at BeGambleAware.org. 18+.