Wednesday 5 August · Draw bias, computed from real settled results — not model output
Kempton
Over the 200 days of results to 14 Jul 2026, horses drawn in the lowest third of stalls at Kempton won 11% of the time, against just 8% for those drawn widest — a 3-point gap across 86 qualifying races (260 runners drawn low, 260 drawn wide).
Based on 86 races with 8+ runners · computed from real settled results, not model output.
Kempton race card →
Yarmouth
No meaningful draw bias at Yarmouth over the 200 days to 14 Jul 2026 — low and high draws have won at close to the same rate across 20 qualifying races.
Based on 20 races with 8+ runners · computed from real settled results, not model output.
Yarmouth race card →
Pontefract
No meaningful draw bias at Pontefract over the 200 days to 14 Jul 2026 — low and high draws have won at close to the same rate across 22 qualifying races.
Based on 22 races with 8+ runners · computed from real settled results, not model output.
General draw-bias knowledge
Up to 8f: Left-handed and undulating with no straight course — every race turns; low (inside) draws hold a well-documented edge, especially from 6f to a mile.
9f+: The turning track keeps a modest low-draw edge at longer trips.
General course knowledge, independent of our own results above — where our own sample is still thin, we show both rather than quietly pick a winner.
Pontefract race card →
Brighton
No meaningful draw bias at Brighton over the 200 days to 14 Jul 2026 — low and high draws have won at close to the same rate across 11 qualifying races.
Based on 11 races with 8+ runners · computed from real settled results, not model output.
General draw-bias knowledge
All distances: Left-handed horseshoe — every race involves the left-hand bend, giving low (inside) draws a modest edge.
General course knowledge, independent of our own results above — where our own sample is still thin, we show both rather than quietly pick a winner.
Brighton race card →
Every finding on this page is arithmetic over real, settled race results — not something our AI model generated. We compute draw-bias win rates per course from 200days of results and publish the number whether or not it's exciting, including courses with no meaningful bias at all.
