Ranked field
Every active runner appears exactly once in model order.
AI-powered racing intelligence · Experimental beta
Understand the race.
Question the odds.
Compare each horse’s estimated chance with the available odds. See the reasoning, spot missing information, and decide what deserves more research.
No guaranteed returns No account needed to explore
Illustrative example No live data shown
Fictional horse · invented numbers
What does the price tell you?
An explanation, not a current tip or model result.
What you compareExample
01
25%Estimated chance
One chance in four
02
4.00Fair odds
The price implied by that estimate
03
5.00Example market odds
A bigger price — worth a closer look
A price gap is a research prompt.
Even if 25% is right, this horse loses 75% of the time.
From numbers to a clearer decision
Know what you’re looking at before you pay.
True Overlay is a research workspace, not a bookmaker or a promise of winning tips. Here is how to read the information it is designed to provide.
01
Compare the field
A ranked view brings estimated chances and market prices together, so you can see where they disagree.
02
Check the reasons
Read the supporting evidence and missing inputs. A price gap matters only if the estimate deserves your trust.
03
Keep a record
Save your research and review it against the official result. A useful process includes the calls that went wrong.
Illustrative example only — not live data
Example Runner: a 25% chance at odds of 5.00
The fictional estimate implies fair decimal odds of 4.00 (1 ÷ 0.25). An invented market price of 5.00 is higher. That is a reason to examine the evidence—not proof of an edge.
- Estimated win chance
- 25%
- Fair decimal odds
- 4.00
- Example market odds
- 5.00
If the true chance were only 18%, fair odds would be about 5.56. The same 5.00 market price would then be too short. The quality of the probability estimate changes the conclusion.
Coverage contract
Built for verified global cards. America next.
Great Britain, Ireland and Hong Kong are the core supported regions. A region becomes live only after its licensed feed and settlement workflow pass verification. North America is the next launch priority, with tote-aware pricing planned from the start.
Supported regions
Product coverage · feed-gated
- GB
Great Britain
Core supported
- IE
Ireland
Core supported
- HK
Hong Kong
Core supported
- INT
International
Selected cards
- US / CA
North America
Launch priority · feed pending
Planned for major US and Canadian circuits with pari-mutuel odds, fractional display and state-aware wagering context.
- AU
Australia
Expansion roadmap
Expansion labels describe product direction, not live availability. A region activates only after its licensed feed and settlement workflow are verified.
Explainable AI predictions
More than a pick.
AI reasoning you can audit.
True Overlay turns verified race data into structured AI predictions—then shows the probability, fair price, evidence and uncertainty behind every call.
Inspect our AI methodology- 01 / 03
Read the whole field
Every runner is ranked. No cherry-picked tip, no missing longshot, no hiding the inconvenient call.
- 02 / 03
Price the probability
We turn estimated win probability into a fair decimal price, then show the market gap in plain sight.
- 03 / 03
Keep the evidence
Source fields, freshness, model version and uncertainty stay attached to every published analysis.
Decision guardrail
Set the rules before the market moves.
Pre-commitment keeps a useful model from becoming an emotional shortcut. Define the bankroll, stake and stop point first.
01
Ring-fence
Only use a separate bankroll you can afford to lose.
02
Size it
Set a fixed maximum stake before opening a card.
03
Stop
Pause when the session limit you chose is reached.
Live AI predictions
Today's intelligence board.
Predictions appear only when the stored card is no more than 30 minutes old and at least 50% of required form and rating fields are present. No qualifying feed means no fabricated picks.
—
Live cards
Integration ready · credentials pending
No pretend races. No placeholder picks.
Connect the database and licensed racing feed to begin importing real cards. Until then, every live-data surface remains honestly unavailable.
View platform status →Analysis standard
A complete record, not a single pick.
Win probability
A deterministic blend of de-vigged market consensus, stored ratings, and bounded AI is normalized to 100%.
Fair decimal price
Calculated deterministically from the normalized probability.
Finish bands
Paid top-two and top-three estimates are deterministically derived from the locked win vector under a disclosed Harville assumption and prospectively scored in the public platform ledger.
Market comparison
Median bookmaker prices are converted to a de-vigged field baseline; best price is kept separately for expected-value arithmetic.
Pace map and shape
AI classifies early run style only from supplied form and comments; code then derives a coverage-gated whole-field pressure diagnosis without changing the forecast.
Field-level evidence
Evidence labels resolve to non-empty source fields for that runner.
Uncertainty
Missing inputs and data completeness remain visible beside the estimate.
Audit identity
Model ID, prompt version, input hash, source time, and generation time are retained.
Settlement
Official results close the record without deleting unsuccessful predictions.
Calibration
Settled runner probabilities feed a public multiclass Brier score, matched market-skill benchmark and forecast-band reliability table.
Operating lifecycle
From source request to official settlement.
- 01
Ingest
Fetch licensed cards and results behind server-only credentials.
- 02
Normalize
Map provider records into stable races, runners, prices, and result identities.
- 03
Generate
Create one structured shared analysis for the current source input.
- 04
Validate
Check identities, evidence, totals, arithmetic, freshness, and completeness.
- 05
Publish
Lock the prediction with its model, prompt, hash, and timestamp.
- 06
Settle
Attach official results while retaining every successful and unsuccessful call.
Institutional standards
Rules that do not move with the result.
A mature analysis product needs operating rules that remain intact after a win, a loss, a correction, or a missing feed.
Source before signal
An analysis is published only when its racecard exists in the normalized store and every cited field can be traced to that record.
Probability before promotion
The product presents hybrid estimated chances, component weights, fair prices, market gaps, and uncertainty. It never turns an estimate into a promised outcome.
The record stays public
Shared pre-race predictions are immutable. Once official results arrive, wins and losses settle into the same public ledger.
Missing data stays missing
Incomplete form, ratings, comments, or prices reduce the displayed completeness score; the system does not invent replacements.
Reference library
Documentation before persuasion.
Methodology
The complete generation and validation contract.
Coverage
Region registry, feed cadence, and freshness rules.
Documentation
Product interpretation and derived API reference.
Platform status
Public configuration readiness at request time.
Security
Architecture, key handling, and trust boundaries.
Disclosures
Provider, AI, price, performance, and affiliate context.
The Blog
From the Paddock Notes
How overlays work, what the model measures, and how to bet an edge without going broke finding out.
Photo: Cauã CardosoForecast Validation7 min read
How to Validate Horse-Racing Predictions Before Trusting Them
A prediction is not evidence of a good model. Here is a simple, repeatable way to test probabilities before confusing a short winning run with skill.
Photo: Keith LukeFundamentals6 min read
What Is an Overlay Bet? Fair Odds, Worked Maths and Uncertainty
An overlay is a comparison between an available price and an uncertain estimate. Follow the arithmetic, then check what would make the apparent advantage disappear.
Photo: Mathew SchwartzModel Inputs6 min read
Sectional Times Explained: Why Closing Speed Beats Final Position
A horse that finishes fourth can still have run a revealing sectional. Timing data can add context that the finishing position alone misses.