Unlucky Last Time.
What did the result miss?
HorseBet.ai is testing whether a bare finishing position can conceal a better performance when a horse is denied a clear run, waits for room, is checked, hampered, stumbled or loses momentum at a significant stage of the race.
Comments find candidates.
We scan latest-run evidence for phrases including waiting for room, not clear run, denied a clear run, short of room, checked, switched, stumbled and other incidents. Keywords start the investigation; they do not create a selection by themselves.
Visual evidence adds judgement.
Where a replay is available, we assess whether the horse was still travelling, how long momentum was interrupted, whether the jockey had to stop riding or switch, and what the horse did once clear. Verdicts are recorded as Strong, Interesting, Unconvincing or Not Viewed.
Class movement is recorded.
Previous class and next-race class are stored separately. At the 50-result review we can compare horses moving up, down or remaining in the same class rather than assuming in advance which group performs best.
Recorded before the result.
Qualifiers and ratings are recorded before racing and retained whether they win or lose. The first formal review is planned after 50 recorded qualifiers, followed by a larger 100+ evidence sample.
Trouble Rating is not a Win Rating.
The rating describes evidence from the previous race; it does not predict that the horse will win next time. HorseBet.ai is a research and intelligence project, not a tipping service.
View the live validation record · Research methodology & responsible gambling