The favourite
What the market's first choice does in the situations the trade talks about.
“When a race falls apart, back the jolly.”
Not in the numbersVerdict: The missing runners are already in the price. With three or more gone the favourite wins no more often than its new odds say — if anything a little less — so there is nothing extra in the jolly once the race has fallen apart.
- The favourite in a race that lost 3 or more declared runners
29.3 %won1.08fair A/E5,012N
- The favourite in a race that lost none
32.6 %won1.09fair A/E267,741N
- The favourite, 1 non-runner
33.5 %won1.06fair A/E14,866N
- The favourite, 2 non-runners
32.8 %won1.06fair A/E5,805N
The favourite in a race that lost 3 or more declared runners won 29.3 % of 5,012 runs at fair A/E 1.08; the favourite in a race that lost none 32.6 % of 267,741 at 1.09. Verdict: not in the numbers.
Rule 4 deductions already take the missing runners out of the price. The question is whether the favourite then wins more often than its new price says.
The last 365 days: the favourite in a race that lost 3 or more declared runners: too few to say (N 890) · the favourite in a race that lost none 34.1 %, fair A/E 1.05 (N 8,479)
Since we first published this, on 25 October 2026: the favourite in a race that lost 3 or more declared runners: no runs yet · the favourite in a race that lost none: too few to say (N 184)
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
Today's non-runners →
“The early favourites are the safe ones.”
True — and pricedVerdict: The early favourite does win more often — but that is the race mix, small fields and the all-weather before lunch, and the price knows it. There is no safety in the hour itself. The price already says so: on this alone the market gives nothing away.
- The favourite, off before 1 pm
37.9 %won1.11fair A/E8,840N
- The favourite, off from 5 pm
31.5 %won1.09fair A/E104,283N
- The favourite, off 1 – 3 pm
34.9 %won1.09fair A/E77,898N
- The favourite, off 3 – 5 pm
31.6 %won1.08fair A/E102,403N
The favourite, off before 1 pm won 37.9 % of 8,840 runs at fair A/E 1.11; the favourite, off from 5 pm 31.5 % of 104,283 at 1.09. Verdict: true — and priced.
Early cards lean to small-field flat and all-weather races, so this compares the favourites the clock serves up, not the clock itself.
The last 365 days: the favourite, off before 1 pm: too few to say (N 541) · the favourite, off from 5 pm 32.4 %, fair A/E 1.06 (N 5,132)
Since we first published this, on 25 October 2026: the favourite, off before 1 pm: too few to say (N 10) · the favourite, off from 5 pm: too few to say (N 41)
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
Today's racecards →
“The stable's second string is the dark horse.”
Not in the numbersVerdict: The dark horse is a myth. The yard's longer-priced runner wins far less often than its stablemate and, worse, runs below even its longer price: the market still gives it too much credit for the yard's name.
- The longer-priced stablemate
4.0 %won15.3 %placed0.87fair A/E151,865N
- The shorter-priced stablemate
11.7 %won31.9 %placed1.01fair A/E150,951N
The longer-priced stablemate won 4.0 % of 151,865 runs at fair A/E 0.87; the shorter-priced stablemate 11.7 % of 150,951 at 1.01. Verdict: not in the numbers.
Counted only in races where one yard ran two or more. The second string is the one the market priced longer, which is how the trade reads the yard's intentions.
The last 365 days: the longer-priced stablemate fair A/E 0.88, won 4.2 % (N 8,304) · the shorter-priced stablemate fair A/E 0.96, won 12.7 % (N 7,854)
Since we first published this, on 25 October 2026: the longer-priced stablemate: too few to say (N 106) · the shorter-priced stablemate: too few to say (N 93)
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
Trainer statistics →
“Odds-on shots are bad value.”
Not in the numbersVerdict: Odds-on shots are the opposite of bad value in fair terms. They win more often than their price implies — the favourite–longshot bias at its purest — so the complaint is about the small return, not the price.
- Odds-on (fair) runners on the flat
63.5 %won1.08fair A/E12,113N
- Odds-on (fair) runners over jumps
65.1 %won1.09fair A/E12,402N
Flat: Odds-on (fair) runners on the flat won 63.5 % of 12,113 runs at fair A/E 1.08 against the fair line of 1.00. Verdict: not in the numbers.
Jumps: Odds-on (fair) runners over jumps won 65.1 % of 12,402 runs at fair A/E 1.09 against the fair line of 1.00. Verdict: not in the numbers.
Odds-on here means a fair winning chance of a half or more once the bookmakers' margin is removed. Compare the fair A/E with the fair line: above 1.00 means these runners won more often than their price said.
The last 365 days: Flat: odds-on (fair) runners on the flat: too few to say (N 783) · the fair line 1.00 — Jumps: odds-on (fair) runners over jumps: too few to say (N 655) · the fair line 1.00
Since we first published this, on 25 October 2026: Flat: odds-on (fair) runners on the flat: too few to say (N 10) · the fair line 1.00 — Jumps: odds-on (fair) runners over jumps: too few to say (N 11) · the fair line 1.00
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
Fair A/E, defined →Odds-on, on the price of a bet →
Going and the seasons
The mud, the changeover and the calendar, measured on the favourite and on the price of the winner.
“Heavy going is a lottery.”
Not in the numbersVerdict: Heavy going is not a lottery in either code: the favourite wins at least as often as on good, and no more races go to a big price.
- The jumps favourite on heavy
36.2 %won63.2 %placed1.12fair A/E15,891N
- The jumps favourite on good
33.8 %won60.2 %placed1.08fair A/E39,869N
- The flat favourite on heavy
28.9 %won57.2 %placed1.03fair A/E5,510N
- The flat favourite on good
30.2 %won59.1 %placed1.07fair A/E34,223N
- Jumps races on heavy
15.5 %won at 10/1 or longer4.4 %won at 20/1 or longer14,736races
- Jumps races on good
16.1 %won at 10/1 or longer4.8 %won at 20/1 or longer36,762races
- Flat races on heavy
18.9 %won at 10/1 or longer5.1 %won at 20/1 or longer5,032races
- Flat races on good
19.8 %won at 10/1 or longer5.8 %won at 20/1 or longer31,327races
Jumps: The jumps favourite on heavy won 36.2 % of 15,891 runs at fair A/E 1.12; the jumps favourite on good 33.8 % of 39,869 at 1.08. Verdict: not in the numbers.
Flat: The flat favourite on heavy won 28.9 % of 5,510 runs at fair A/E 1.03; the flat favourite on good 30.2 % of 34,223 at 1.07. Verdict: not in the numbers.
A lottery would show in two places: the favourite winning less often, and more races going to a horse at 10/1 or longer. Both are read.
The last 365 days: Jumps: the jumps favourite on heavy: too few to say (N 584) · the jumps favourite on good 34.7 %, fair A/E 1.04 (N 2,150) — Flat: the flat favourite on heavy: too few to say (N 199) · the flat favourite on good 33.0 %, fair A/E 1.08 (N 1,978)
Since we first published this, on 25 October 2026: Jumps: the jumps favourite on heavy: too few to say (N 16) · the jumps favourite on good: too few to say (N 22) — Flat: the flat favourite on heavy: too few to say (N 6) · the flat favourite on good: no runs yet
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
Racecourse statistics →The favourite by going →
“Only back the short ones in the mud.”
Not in the numbersVerdict: Backwards. On heavy it is the bigger-priced favourite that runs above its price, not the odds-on one, which behaves on heavy exactly as it does everywhere else.
- The favourite at 3/1 or longer (fair) on heavy
21.6 %won1.14fair A/E8,332N
- The odds-on favourite on heavy
64.8 %won1.09fair A/E2,300N
- The favourite at 3/1 or longer on other going
21.2 %won1.10fair A/E94,940N
- The odds-on favourite on other going
64.2 %won1.08fair A/E17,972N
The favourite at 3/1 or longer (fair) on heavy won 21.6 % of 8,332 runs at fair A/E 1.14; the odds-on favourite on heavy 64.8 % of 2,300 at 1.09. Verdict: not in the numbers.
The favourite's own price band is the fair one, read from the whole book with the margin removed. The same bands on other going are printed beside the heavy rows.
The last 365 days: the favourite at 3/1 or longer (fair) on heavy: too few to say (N 277) · the odds-on favourite on heavy: too few to say (N 98)
Since we first published this, on 25 October 2026: the favourite at 3/1 or longer (fair) on heavy: too few to say (N 9) · the odds-on favourite on heavy: too few to say (N 1)
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
Fair A/E, defined →The favourite by going →
“Racing gets hard to read around the autumn changeover.”
Not in the numbersVerdict: The favourite in the autumn changeover window wins as often as in summer in both codes: the changeover is a story, not a season.
- The flat favourite, 15 October – 15 November
29.7 %won59.9 %placed1.09fair A/E12,714N
- The flat favourite, May – September
31.6 %won59.8 %placed1.09fair A/E108,379N
- The jumps favourite, 15 October – 15 November
34.9 %won61.5 %placed1.10fair A/E13,024N
- The jumps favourite, May – September
33.0 %won60.6 %placed1.09fair A/E36,034N
- Flat races, 15 Oct – 15 Nov
21.6 %won at 10/1 or longer6.4 %won at 20/1 or longer11,607races
- Flat races, May – Sep
18.0 %won at 10/1 or longer5.0 %won at 20/1 or longer99,125races
- Jumps races, 15 Oct – 15 Nov
15.3 %won at 10/1 or longer4.3 %won at 20/1 or longer11,995races
- Jumps races, May – Sep
17.1 %won at 10/1 or longer4.8 %won at 20/1 or longer33,160races
Flat: The flat favourite, 15 October – 15 November won 29.7 % of 12,714 runs at fair A/E 1.09; the flat favourite, May – September 31.6 % of 108,379 at 1.09. Verdict: not in the numbers.
Jumps: The jumps favourite, 15 October – 15 November won 34.9 % of 13,024 runs at fair A/E 1.10; the jumps favourite, May – September 33.0 % of 36,034 at 1.09. Verdict: not in the numbers.
The two codes answer separately: the big autumn flat handicaps run in larger fields than the summer's, and the early jumps season is its own thing.
The last 365 days: Flat: the flat favourite, 15 October – 15 November: too few to say (N 588) · the flat favourite, May – September 33.3 %, fair A/E 1.05 (N 5,037) — Jumps: the jumps favourite, 15 October – 15 November: too few to say (N 598) · the jumps favourite, May – September 33.1 %, fair A/E 1.02 (N 1,536)
Since we first published this, on 25 October 2026: Flat: the flat favourite, 15 October – 15 November: too few to say (N 101) · the flat favourite, May – September: no runs yet — Jumps: the jumps favourite, 15 October – 15 November: too few to say (N 83) · the jumps favourite, May – September: no runs yet
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
Today's racecards →The favourite by season →
“A jumps horse needs the run after its summer break.”
Not in the numbersVerdict: "Needs the run" is not in the numbers. The autumn jumps favourite off a summer break wins as often as one that raced within the last two months — the yards that send a fresh favourite out send it fit.
- The autumn jumps favourite off 150 days or more
36.8 %won61.9 %placed1.10fair A/E4,465N
- The autumn jumps favourite raced within 60 days
34.1 %won61.4 %placed1.11fair A/E7,286N
- The autumn jumps favourite off 60 – 149 days
too few to say478N
The autumn jumps favourite off 150 days or more won 36.8 % of 4,465 runs at fair A/E 1.10; the autumn jumps favourite raced within 60 days 34.1 % of 7,286 at 1.11. Verdict: not in the numbers.
Favourites in the autumn changeover window only, split by the length of their break. "Needs the run" predicts the fresh ones win less.
The last 365 days: the autumn jumps favourite off 150 days or more: too few to say (N 243) · the autumn jumps favourite raced within 60 days: too few to say (N 289)
Since we first published this, on 25 October 2026: the autumn jumps favourite off 150 days or more: too few to say (N 32) · the autumn jumps favourite raced within 60 days: too few to say (N 42)
Every GB and Irish race from 15 October 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
“The spring changeover is the easier one.”
Not in the numbersVerdict: Not the easier one. The flat favourite in the spring window wins no more often than in the autumn, and the market reads it worse there than in any other season.
- The flat favourite, 15 March – 30 April
30.8 %won59.4 %placed1.05fair A/E17,979N
- The flat favourite, 15 October – 15 November
29.7 %won59.9 %placed1.09fair A/E12,714N
- The jumps favourite, 15 March – 30 April
33.2 %won59.5 %placed1.09fair A/E18,083N
- The jumps favourite, 15 October – 15 November
34.9 %won61.5 %placed1.10fair A/E13,024N
The flat favourite, 15 March – 30 April won 30.8 % of 17,979 runs at fair A/E 1.05; the flat favourite, 15 October – 15 November 29.7 % of 12,714 at 1.09. Verdict: not in the numbers.
"Easier to read" is taken to mean the favourite wins more often. The fair A/E beside each row says how well the market read it.
The last 365 days: the flat favourite, 15 March – 30 April: too few to say (N 942) · the flat favourite, 15 October – 15 November: too few to say (N 588)
Since we first published this, on 25 October 2026: the flat favourite, 15 March – 30 April: no runs yet · the flat favourite, 15 October – 15 November: too few to say (N 101)
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
The favourite by season →
“Back the three-year-olds against the older horses in the autumn.”
Half trueVerdict: It depends on the month. The weight-for-age scale is designed to level the generations and in most months the market prices it to the pound; where the three-year-olds run above their price it is one month, not the autumn.
- Three-year-olds against older horses, September
10.5 %won1.02fair A/E38,273N
- Older horses against three-year-olds, September
8.3 %won0.99fair A/E64,226N
- Three-year-olds against older horses, October
9.6 %won1.03fair A/E36,430N
- Older horses against three-year-olds, October
7.8 %won0.98fair A/E59,506N
- Three-year-olds against older horses, November
9.9 %won1.02fair A/E17,528N
- Older horses against three-year-olds, November
8.5 %won0.99fair A/E33,077N
September: Three-year-olds against older horses, September won 10.5 % of 38,273 runs at fair A/E 1.02; older horses against three-year-olds, September 8.3 % of 64,226 at 0.99. Verdict: not in the numbers.
October: Three-year-olds against older horses, October won 9.6 % of 36,430 runs at fair A/E 1.03; older horses against three-year-olds, October 7.8 % of 59,506 at 0.98. Verdict: true.
November: Three-year-olds against older horses, November won 9.9 % of 17,528 runs at fair A/E 1.02; older horses against three-year-olds, November 8.5 % of 33,077 at 0.99. Verdict: not in the numbers.
Counted only in flat races with both three-year-olds and older horses in the field. The weight-for-age scale is designed to level them; the fair A/E says whether the market has.
The last 365 days: September: three-year-olds against older horses, September fair A/E 1.03, won 10.1 % (N 2,043) · older horses against three-year-olds, September fair A/E 0.98, won 8.6 % (N 3,172) — October: three-year-olds against older horses, October fair A/E 0.97, won 9.4 % (N 1,893) · older horses against three-year-olds, October fair A/E 1.03, won 8.8 % (N 2,994) — November: three-year-olds against older horses, November: too few to say (N 918) · older horses against three-year-olds, November fair A/E 1.04, won 9.3 % (N 1,606)
Since we first published this, on 25 October 2026: October: three-year-olds against older horses, October: too few to say (N 236) · older horses against three-year-olds, October: too few to say (N 434)
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
Weight, claims, age and sex
The pounds on the back, the allowance, the veterans and the fillies.
“Top weight can't win a big handicap.”
Not in the numbersVerdict: The top weight is the horse to fear, not to dismiss. In a big-field handicap it wins more often than the bottom weight, not less, and is fairly priced doing it — it carries the most because it is rated the best, and the mark wins more than the weight costs.
- The top weight in a handicap of 14 or more
8.3 %won1.03fair A/E35,151N
- The bottom weight in a handicap of 14 or more
4.3 %won0.97fair A/E34,038N
- The top weight, 9 runners or fewer
18.4 %won1.01fair A/E90,161N
- The top weight, 10 – 13 runners
11.6 %won1.02fair A/E70,316N
- Neither top nor bottom, 14 or more
6.2 %won1.00fair A/E370,607N
The top weight in a handicap of 14 or more won 8.3 % of 35,151 runs at fair A/E 1.03; the bottom weight in a handicap of 14 or more 4.3 % of 34,038 at 0.97. Verdict: not in the numbers.
The top weight is usually the highest-rated horse in the race. The question is whether the weight it carries for that costs it more than the mark earns.
The last 365 days: the top weight in a handicap of 14 or more 8.3 %, fair A/E 1.03 (N 1,442) · the bottom weight in a handicap of 14 or more 4.2 %, fair A/E 0.91 (N 1,301)
Since we first published this, on 25 October 2026: the top weight in a handicap of 14 or more: too few to say (N 21) · the bottom weight in a handicap of 14 or more: too few to say (N 14)
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
Off a winning mark, today →
“Weight tells in the mud.”
Half trueVerdict: Half true: weight tells in the mud on the flat only. The flat handicap top weight runs below its price on soft or heavy and above it on faster ground; over jumps the top weight runs to its price whatever the going.
- The flat handicap top weight on soft or heavy
12.4 %won0.97fair A/E15,125N
- The flat handicap top weight on good or faster
14.8 %won1.03fair A/E103,365N
- The jumps handicap top weight on soft or heavy
13.4 %won1.00fair A/E24,033N
- The jumps handicap top weight on good or faster
14.2 %won1.00fair A/E29,383N
Flat: The flat handicap top weight on soft or heavy won 12.4 % of 15,125 runs at fair A/E 0.97; the flat handicap top weight on good or faster 14.8 % of 103,365 at 1.03. Verdict: true.
Jumps: The jumps handicap top weight on soft or heavy won 13.4 % of 24,033 runs at fair A/E 1.00; the jumps handicap top weight on good or faster 14.2 % of 29,383 at 1.00. Verdict: not in the numbers.
The top weight in a handicap is the test case: if weight tells more in the mud, the top weight should run below its price there and not elsewhere.
The last 365 days: Flat: the flat handicap top weight on soft or heavy: too few to say (N 664) · the flat handicap top weight on good or faster fair A/E 1.00, won 15.3 % (N 5,947) — Jumps: the jumps handicap top weight on soft or heavy fair A/E 0.98, won 13.6 % (N 1,009) · the jumps handicap top weight on good or faster fair A/E 0.96, won 16.2 % (N 1,578)
Since we first published this, on 25 October 2026: Flat: the flat handicap top weight on soft or heavy: too few to say (N 43) · the flat handicap top weight on good or faster: too few to say (N 50) — Jumps: the jumps handicap top weight on soft or heavy: too few to say (N 10) · the jumps handicap top weight on good or faster: too few to say (N 18)
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
“A 7 lb claimer is worth having.”
Not in the numbersVerdict: The allowance is priced to the pound. A horse with a claimer up wins at the rate its price says — the bigger the claim, if anything, the less — so the pounds off are not a free gift.
- Ridden by a 7 lb or heavier claimer
7.0 %won0.97fair A/E271,818N
- No claim
11.0 %won1.01fair A/E1,895,412N
- Ridden by a 3 lb claimer
8.6 %won1.00fair A/E255,289N
- Ridden by a 5 lb claimer
8.3 %won0.99fair A/E237,567N
Ridden by a 7 lb or heavier claimer won 7.0 % of 271,818 runs at fair A/E 0.97; no claim 11.0 % of 1,895,412 at 1.01. Verdict: not in the numbers.
The allowance is taken off the horse's weight. If the trade is right that an inexperienced rider costs less than the pounds are worth, the claimers' fair A/E should sit above the rest.
The last 365 days: ridden by a 7 lb or heavier claimer fair A/E 0.95, won 7.4 % (N 10,398) · no claim fair A/E 1.01, won 11.6 % (N 91,101)
Since we first published this, on 25 October 2026: ridden by a 7 lb or heavier claimer: too few to say (N 119) · no claim fair A/E 0.99, won 10.9 % (N 1,311)
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
Jockey statistics →
“The old boys are still worth backing.”
Not in the numbersVerdict: The old boys are over-backed. Horses aged ten and over run below their price — the market pays for the reputation, and the reputation is worth a little less than it costs.
- Aged ten and over
7.6 %won0.94fair A/E117,950N
- Aged five or six
10.3 %won1.02fair A/E759,962N
- Aged seven to nine
9.7 %won0.99fair A/E547,878N
Aged ten and over won 7.6 % of 117,950 runs at fair A/E 0.94 against the fair line of 1.00. Verdict: not in the numbers.
Every runner by age, against the fair line. Above 1.00 the veterans are winning more than their price says; below it, less.
The last 365 days: aged ten and over fair A/E 1.00, won 8.4 % (N 3,922) · the fair line 1.00
Since we first published this, on 25 October 2026: aged ten and over: too few to say (N 40) · the fair line 1.00
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
Horse statistics →
“Fillies can't beat the colts.”
Not in the numbersVerdict: Fillies and mares hold their own against males on the flat: they win less often but the allowance covers it closely enough that the market gap is within our threshold. The belief is not in the numbers.
- Fillies and mares against males
8.5 %won0.98fair A/E382,637N
- Males against fillies and mares
10.4 %won1.01fair A/E837,216N
Fillies and mares against males won 8.5 % of 382,637 runs at fair A/E 0.98; males against fillies and mares 10.4 % of 837,216 at 1.01. Verdict: not in the numbers.
Counted only in flat races with both sexes in the field, where the fillies' allowance applies. The belief says the allowance is not enough.
The last 365 days: fillies and mares against males fair A/E 0.98, won 9.0 % (N 17,249) · males against fillies and mares fair A/E 1.01, won 10.9 % (N 37,997)
Since we first published this, on 25 October 2026: fillies and mares against males: too few to say (N 284) · males against fillies and mares: too few to say (N 568)
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
“A horse back out within a few days is being rushed.”
Not in the numbersVerdict: The opposite of rushed. A horse back out within three days runs above its price in both codes — a yard brings a horse straight back because it came out of the last race well, and the market under-reads that.
- Jumps, back within 3 days
14.1 %won1.15fair A/E5,670N
- Jumps, 8 – 21 days since the last run
10.1 %won0.98fair A/E307,181N
- Flat, back within 3 days
14.4 %won1.09fair A/E19,658N
- Flat, 8 – 21 days since the last run
10.6 %won0.99fair A/E637,689N
- Jumps, 4 – 7 days
12.4 %won1.09fair A/E30,626N
- Flat, 4 – 7 days
12.9 %won1.05fair A/E97,858N
Jumps: Jumps, back within 3 days won 14.1 % of 5,670 runs at fair A/E 1.15; jumps, 8 – 21 days since the last run 10.1 % of 307,181 at 0.98. Verdict: not in the numbers.
Flat: Flat, back within 3 days won 14.4 % of 19,658 runs at fair A/E 1.09; flat, 8 – 21 days since the last run 10.6 % of 637,689 at 0.99. Verdict: not in the numbers.
Days since the last run, from our own sequencing of each horse's record. A rushed horse should run below its price; a horse brought back because it is well should run above it.
The last 365 days: Jumps: jumps, back within 3 days: too few to say (N 109) · jumps, 8 – 21 days since the last run fair A/E 1.00, won 10.8 % (N 10,600) — Flat: flat, back within 3 days: too few to say (N 579) · flat, 8 – 21 days since the last run fair A/E 0.99, won 11.0 % (N 28,273)
Since we first published this, on 25 October 2026: Jumps: jumps, back within 3 days: no runs yet · jumps, 8 – 21 days since the last run: too few to say (N 143) — Flat: flat, back within 3 days: too few to say (N 6) · flat, 8 – 21 days since the last run: too few to say (N 430)
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
Headgear, the course, the yard move, the class and the trip — the flags our sequencer sets before each run.
“First-time blinkers work wonders.”
Not in the numbersVerdict: Blinkers are not a magic switch. First-time headgear runs no better than its price in either code; the change is already in the odds, and in the record it is slightly cold.
- First-time headgear, flat
8.2 %won0.95fair A/E63,966N
- No first-time headgear, flat
10.3 %won1.00fair A/E1,497,766N
- First-time headgear, jumps
8.0 %won0.98fair A/E44,646N
- No first-time headgear, jumps
10.3 %won1.00fair A/E1,007,839N
Flat: First-time headgear, flat won 8.2 % of 63,966 runs at fair A/E 0.95; no first-time headgear, flat 10.3 % of 1,497,766 at 1.00. Verdict: not in the numbers.
Jumps: First-time headgear, jumps won 8.0 % of 44,646 runs at fair A/E 0.98; no first-time headgear, jumps 10.3 % of 1,007,839 at 1.00. Verdict: not in the numbers.
First-time headgear is our own flag, sequenced from each horse's record (not the racecard's symbol, which is sometimes absent). Today's first-timers are listed on their own page.
The last 365 days: Flat: first-time headgear, flat fair A/E 1.00, won 9.1 % (N 3,110) · no first-time headgear, flat fair A/E 1.00, won 10.8 % (N 70,894) — Jumps: first-time headgear, jumps fair A/E 0.98, won 9.6 % (N 2,151) · no first-time headgear, jumps fair A/E 1.00, won 11.2 % (N 43,085)
Since we first published this, on 25 October 2026: Flat: first-time headgear, flat: too few to say (N 36) · no first-time headgear, flat: too few to say (N 994) — Jumps: first-time headgear, jumps: too few to say (N 25) · no first-time headgear, jumps: too few to say (N 632)
Every GB and Irish race from 3 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
First-time headgear, today →
“Horses for courses.”
True — and pricedVerdict: Horses for courses holds on the strike rate: the course-and-distance winner wins more often than a horse on its first visit. The price already says so: on this alone the market gives nothing away.
- A course-and-distance winner
13.0 %won1.00fair A/E269,365N
- First run at the course
10.0 %won1.00fair A/E1,486,947N
- A course winner, not over the distance
12.2 %won1.03fair A/E103,640N
- Ran here before, never won
9.2 %won1.00fair A/E800,134N
A course-and-distance winner won 13.0 % of 269,365 runs at fair A/E 1.00; first run at the course 10.0 % of 1,486,947 at 1.00. Verdict: true — and priced.
Course form as our sequencer records it before each run. The belief is that the course winner wins more; the fair A/E says whether the market has already paid for it.
The last 365 days: a course-and-distance winner 12.9 %, fair A/E 0.98 (N 14,478) · first run at the course 10.8 %, fair A/E 1.00 (N 64,443)
Since we first published this, on 25 October 2026: a course-and-distance winner: too few to say (N 195) · first run at the course: too few to say (N 957)
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
Racecourse statistics →
“A change of yard works the oracle.”
Not in the numbersVerdict: A change of yard works no oracle. The first run for a new yard is, in the record, a run below the price: the market gives the move more credit than it earns.
- First run for a new yard
8.4 %won1.02fair A/E102,521N
- Same yard as the last run
10.5 %won1.00fair A/E2,348,840N
First run for a new yard won 8.4 % of 102,521 runs at fair A/E 1.02; same yard as the last run 10.5 % of 2,348,840 at 1.00. Verdict: not in the numbers.
The first run after a move between yards, from our own sequencing. A move that works should show as runs above the price.
The last 365 days: first run for a new yard fair A/E 1.02, won 8.9 % (N 4,602) · same yard as the last run fair A/E 1.00, won 11.2 % (N 106,286)
Since we first published this, on 25 October 2026: first run for a new yard: too few to say (N 81) · same yard as the last run fair A/E 1.02, won 10.8 % (N 1,460)
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
Trainer statistics →
“Dropped in class, it's thrown in.”
Not in the numbersVerdict: "Thrown in" is already in the price. A horse down in class wins more often, as it should, but runs to its odds, not above them: the drop is the first thing the market sees.
- Down in class from the last run
12.1 %won1.01fair A/E416,873N
- Same class as the last run
11.1 %won1.00fair A/E1,012,335N
- Up in class from the last run
10.0 %won0.99fair A/E401,421N
Down in class from the last run won 12.1 % of 416,873 runs at fair A/E 1.01; same class as the last run 11.1 % of 1,012,335 at 1.00. Verdict: not in the numbers.
The class move is read from each race's class against the horse's last. "Thrown in" predicts runs above the price.
The last 365 days: down in class from the last run fair A/E 1.01, won 12.8 % (N 15,872) · same class as the last run fair A/E 1.01, won 12.1 % (N 47,972)
Since we first published this, on 25 October 2026: down in class from the last run: too few to say (N 267) · same class as the last run: too few to say (N 734)
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
“Stepping up in trip will bring out the best in it.”
Not in the numbersVerdict: The step up in trip is not an edge. A horse going two furlongs or more further than last time runs to its price, not above it: whatever the trip will bring out, the market has allowed for.
- Two furlongs or more further than the last run
8.8 %won0.99fair A/E314,383N
- The same distance as the last run
11.4 %won1.01fair A/E778,782N
- Two furlongs or more shorter than the last run
9.1 %won0.99fair A/E227,142N
Two furlongs or more further than the last run won 8.8 % of 314,383 runs at fair A/E 0.99; the same distance as the last run 11.4 % of 778,782 at 1.01. Verdict: not in the numbers.
The move in distance from the last run, in furlongs. The belief is about the step up; the step down is printed beside it.
The last 365 days: two furlongs or more further than the last run fair A/E 1.01, won 10.3 % (N 12,231) · the same distance as the last run fair A/E 1.00, won 11.9 % (N 35,646)
Since we first published this, on 25 October 2026: two furlongs or more further than the last run: too few to say (N 148) · the same distance as the last run: too few to say (N 478)
Every GB and Irish race from 2 January 2005 to 29 October 2026. N is runs; fair A/E over runs with a usable starting price.
“'Good' ground means a fast time.”
Not in the numbersVerdict: "Good" is slow on the clock. Only a small share of flat races on official good ground finish inside the course standard — standards are near-record times, so running inside one is rare on any going. The useful reading is how far outside it a course usually runs, which is on each course's page.
Of 29,317 flat races run on official 'Good' going, 3.2 % finished inside the course standard time; on average the winner was 0.46 seconds per furlong outside it.
Course standards are near-record times, so running inside one is rare on any going. The course-by-course spread is the useful part, on each course's statistics page.
Since we first published this, on 25 October 2026: the first full day's results are counted tomorrow.
Every GB and Irish race from 19 March 2005 to 4 November 2025. N is runs; fair A/E over runs with a usable starting price.
Racecourse statistics →KTB Speed, explained →
The horse's last run
What the previous run says about the next one, read as it stood the day before the race.
“Winners keep winning.”
True — and the market has not caught upAdded 28 OctoberVerdict: Winners do keep winning. A horse that won its last two wins far more often than one that did not win last time, and the hat-trick bid is one of the few streaks the market under-prices. And the market has not caught up: this group has been running above its price.
- Won its last two runs
22.0 %won1.04fair A/E46,637N
- Did not win last time
9.2 %won0.99fair A/E2,399,668N
- Won last time only
17.8 %won1.04fair A/E213,781N
Won its last two runs won 22.0 % of 46,637 runs at fair A/E 1.04; did not win last time 9.2 % of 2,399,668 at 0.99. Verdict: true — and the market has not caught up.
The last two runs as they stood the day before the race. A horse bidding for a hat-trick is in the first row.
The last 365 days: won its last two runs 20.1 %, fair A/E 0.96 (N 2,309) · did not win last time 9.9 %, fair A/E 1.00 (N 106,649)
Since we first published this, on 28 October 2026: won its last two runs: too few to say (N 5) · did not win last time: too few to say (N 678)
Every GB and Irish race from 2 January 2005 to 29 October 2026; each record read as it stood the day before the race. N is runs; fair A/E over runs with a usable starting price.
“It needs to get back down to its last winning mark.”
Not in the numbersAdded 28 OctoberVerdict: Backwards. A handicapper running above its last winning mark wins more often than one that has dropped below it, not less: being above the mark is the sign of a progressive horse the handicapper is chasing, not a bar.
- Running 5 lb or more above its best winning mark
12.1 %won1.02fair A/E306,914N
- Running 4 lb or more below it
8.9 %won0.99fair A/E246,579N
- Running 1 – 4 lb above its best winning mark
10.7 %won0.99fair A/E208,485N
- Running 0 – 3 lb below it
10.9 %won0.99fair A/E130,207N
Running 5 lb or more above its best winning mark won 12.1 % of 306,914 runs at fair A/E 1.02; running 4 lb or more below it 8.9 % of 246,579 at 0.99. Verdict: not in the numbers.
Handicaps only; the best mark the horse had ever won off, as it stood before the race. The belief says the horse above its mark is held.
The last 365 days: running 5 lb or more above its best winning mark 13.0 %, fair A/E 1.01 (N 13,200) · running 4 lb or more below it 9.3 %, fair A/E 0.97 (N 14,063)
Since we first published this, on 28 October 2026: running 5 lb or more above its best winning mark: too few to say (N 71) · running 4 lb or more below it: too few to say (N 111)
Every GB and Irish race from 2 January 2005 to 29 October 2026; each record read as it stood the day before the race. N is runs; fair A/E over runs with a usable starting price.
Off a winning mark, today →
“Beaten at odds-on, it can't be trusted.”
Not in the numbersAdded 28 OctoberVerdict: The opposite. A horse beaten at odds-on last time bounces back next time more often than one beaten at a longer price, and runs above its new odds: punters desert it and the record says they should not.
- Beaten at odds-on (fair) last time
25.1 %won1.09fair A/E8,134N
- Beaten at 3/1 or shorter (fair) last time
19.0 %won1.04fair A/E112,629N
- Won at odds-on last time
27.5 %won1.06fair A/E15,107N
Beaten at odds-on (fair) last time won 25.1 % of 8,134 runs at fair A/E 1.09; beaten at 3/1 or shorter (fair) last time 19.0 % of 112,629 at 1.04. Verdict: not in the numbers.
The previous run's fair price and result. If punters desert the beaten odds-on shot, its next-time fair A/E should sit above the horse beaten at a longer price.
The last 365 days: beaten at odds-on (fair) last time: too few to say (N 455) · beaten at 3/1 or shorter (fair) last time fair A/E 1.07, won 20.2 % (N 6,266)
Since we first published this, on 28 October 2026: beaten at odds-on (fair) last time: too few to say (N 3) · beaten at 3/1 or shorter (fair) last time: too few to say (N 25)
Every GB and Irish race from 2 January 2005 to 29 October 2026; each record read as it stood the day before the race. N is runs; fair A/E over runs with a usable starting price.
“Well beaten on debut, it will improve for the run.”
Not in the numbersAdded 28 OctoberVerdict: A well-beaten debutant does not improve for the run — or not as much as the market hopes. Seventh or worse on debut is the worst second-run bet in the record, running well below its price.
- Seventh or worse on debut
3.5 %won0.87fair A/E85,627N
- Fourth to sixth on debut
9.9 %won0.97fair A/E53,235N
- Second on debut
25.0 %won1.02fair A/E14,950N
- Won on debut
23.0 %won1.07fair A/E14,279N
Seventh or worse on debut won 3.5 % of 85,627 runs at fair A/E 0.87; fourth to sixth on debut 9.9 % of 53,235 at 0.97. Verdict: not in the numbers.
The second run of a career, split by the debut finish. "Improve for the run" predicts the well-beaten debutant runs above its price next time.
The last 365 days: seventh or worse on debut fair A/E 0.83, won 2.6 % (N 2,880) · fourth to sixth on debut fair A/E 1.01, won 10.1 % (N 2,503)
Since we first published this, on 28 October 2026: seventh or worse on debut: too few to say (N 29) · fourth to sixth on debut: too few to say (N 18)
Every GB and Irish race from 3 January 2005 to 29 October 2026; each record read as it stood the day before the race. N is runs; fair A/E over runs with a usable starting price.
“Beaten out of sight last time, it's overpriced now.”
Not in the numbersAdded 28 OctoberVerdict: Not overpriced. A horse beaten twenty lengths or more last time still runs below its price next time: the market forgives the heavy defeat a little too readily.
- Beaten 20 lengths or more last time
5.2 %won0.94fair A/E96,301N
- Beaten half a length or less last time
10.9 %won1.01fair A/E517,415N
- Beaten 3 – 10 lengths last time
8.7 %won0.99fair A/E467,262N
Beaten 20 lengths or more last time won 5.2 % of 96,301 runs at fair A/E 0.94; beaten half a length or less last time 10.9 % of 517,415 at 1.01. Verdict: not in the numbers.
How far the horse was beaten on its previous run. The belief says the market over-punishes a heavy defeat.
The last 365 days: beaten 20 lengths or more last time fair A/E 1.02, won 5.7 % (N 3,320) · beaten half a length or less last time fair A/E 1.01, won 11.6 % (N 25,198)
Since we first published this, on 28 October 2026: beaten 20 lengths or more last time: too few to say (N 27) · beaten half a length or less last time: too few to say (N 167)
Every GB and Irish race from 2 January 2005 to 29 October 2026; each record read as it stood the day before the race. N is runs; fair A/E over runs with a usable starting price.
“The unexposed handicap debutant is the one to be on.”
Not in the numbersAdded 28 OctoberVerdict: "Unexposed" is priced. The lightly-raced horse in its first handicap wins more often than the handicap regular, as the market expects, and runs to its odds: there is nothing left over for knowing it is unexposed.
- First handicap after one or two runs
13.6 %won1.01fair A/E12,880N
- Has run in a handicap before
10.0 %won1.00fair A/E1,452,314N
- First handicap after three or four runs
9.1 %won0.99fair A/E83,892N
- First handicap after five or more runs
8.2 %won0.99fair A/E38,596N
First handicap after one or two runs won 13.6 % of 12,880 runs at fair A/E 1.01; has run in a handicap before 10.0 % of 1,452,314 at 1.00. Verdict: not in the numbers.
Handicaps only, by whether the horse had run in one before and how many runs it had. "Unexposed" predicts the market has not caught up with the lightly-raced one.
The last 365 days: first handicap after one or two runs: too few to say (N 404) · has run in a handicap before fair A/E 1.00, won 10.7 % (N 72,173)
Since we first published this, on 28 October 2026: first handicap after one or two runs: too few to say (N 2) · has run in a handicap before: too few to say (N 465)
Every GB and Irish race from 3 January 2005 to 29 October 2026; each record read as it stood the day before the race. N is runs; fair A/E over runs with a usable starting price.
“It's won on the ground — that's the key.”
True — and pricedAdded 28 OctoberVerdict: It wins more often on ground it has won on before — but the market knows the form book too, so the horse's own going record is already in its price. The ground signal that is not priced is the sire's, on the horse's page. The price already says so: on this alone the market gives nothing away.
- Has won on today's going before
12.2 %won1.01fair A/E778,019N
- First run on today's going
9.2 %won0.99fair A/E728,972N
- Ran on it before, never won
9.3 %won1.00fair A/E1,153,095N
Has won on today's going before won 12.2 % of 778,019 runs at fair A/E 1.01; first run on today's going 9.2 % of 728,972 at 0.99. Verdict: true — and priced.
The horse's own record on the going bucket, as it stood before the race. "The key" predicts it wins more; the fair A/E says whether the market already knew.
The last 365 days: has won on today's going before 12.3 %, fair A/E 0.98 (N 40,854) · first run on today's going 10.3 %, fair A/E 1.00 (N 28,271)
Since we first published this, on 28 October 2026: has won on today's going before: too few to say (N 242) · first run on today's going: too few to say (N 203)
Every GB and Irish race from 2 January 2005 to 29 October 2026; each record read as it stood the day before the race. N is runs; fair A/E over runs with a usable starting price.
The yard, the rider and the sire
The records behind the runner — the yard's, the rider's and the sire's — as they stood the day before the race.
“A horse off six months — it depends on the yard.”
True — and the market has not caught upAdded 28 OctoberVerdict: It depends on the yard. A horse off six months from a yard that wins with its fresh ones wins several times as often as one from a yard that does not, and the market has not caught up: it prices the days off, not the yard's record with them. And the market has not caught up: this group has been running above its price.
- Off 180+ days, from a yard that wins 15 % or more with fresh horses
19.0 %won1.09fair A/E34,207N
- Off 180+ days, from a yard under 8 % with fresh horses
5.2 %won0.90fair A/E66,322N
- Off 180+ days, from a yard at 8 – 14 % with fresh horses
11.0 %won1.02fair A/E53,575N
Off 180+ days, from a yard that wins 15 % or more with fresh horses won 19.0 % of 34,207 runs at fair A/E 1.09; off 180+ days, from a yard under 8 % with fresh horses 5.2 % of 66,322 at 0.90. Verdict: true — and the market has not caught up.
Horses off 180 days or more, by their yard's record with earlier fresh runners as it stood the day before. The yard's record is the test, not the days.
The last 365 days: off 180+ days, from a yard that wins 15 % or more with fresh horses 18.4 %, fair A/E 1.01 (N 2,042) · off 180+ days, from a yard under 8 % with fresh horses 4.6 %, fair A/E 0.84 (N 3,256)
Since we first published this, on 28 October 2026: off 180+ days, from a yard that wins 15 % or more with fresh horses: too few to say (N 37) · off 180+ days, from a yard under 8 % with fresh horses: too few to say (N 23)
Every GB and Irish race from 7 July 2005 to 29 October 2026; each record read as it stood the day before the race. N is runs; fair A/E over runs with a usable starting price.
Trainer statistics →
“A big-name booking on a quiet yard's horse means something.”
Not in the numbersAdded 28 OctoberVerdict: The booking on its own means nothing. A top rider for a quiet yard's horse runs exactly to its price: the market reads the booking too, and the horse is still the quiet yard's horse. The rider adds to a hot yard, not to a cold one.
- A top rider (18 % or more over 90 days) for a quiet yard (under 8 %)
11.9 %won1.00fair A/E25,787N
- A top rider for a hot yard (18 % or more)
22.8 %won1.06fair A/E85,406N
- A quiet rider (under 8 %) for a hot yard
11.7 %won1.00fair A/E29,632N
A top rider (18 % or more over 90 days) for a quiet yard (under 8 %) won 11.9 % of 25,787 runs at fair A/E 1.00 against the fair line of 1.00. Verdict: not in the numbers.
The rider's and the yard's 90-day form crossed, as they stood the day before. If the booking itself is a sign, the quiet yard's horse with the top rider should run above its price.
The last 365 days: a top rider (18 % or more over 90 days) for a quiet yard (under 8 %) fair A/E 0.97, won 13.0 % (N 1,164) · the fair line 1.00
Since we first published this, on 28 October 2026: a top rider (18 % or more over 90 days) for a quiet yard (under 8 %): too few to say (N 2) · the fair line 1.00
Every GB and Irish race from 2 January 2005 to 29 October 2026; each record read as it stood the day before the race. N is runs; fair A/E over runs with a usable starting price.
Jockey statistics →
“Put a top jockey on the favourite and it can't lose.”
True — and pricedAdded 28 OctoberVerdict: A favourite ridden by a rider with a strong record on favourites does win more often — and the market has priced the rider in: both favourites run to their odds, so the top jockey is not a reason on his own. The price already says so: on this alone the market gives nothing away.
- The favourite, ridden by a rider who wins 36 % or more on favourites
38.4 %won1.10fair A/E39,618N
- The favourite, ridden by a rider under 30 % on favourites
30.9 %won1.09fair A/E46,349N
- The favourite, ridden by a rider at 30 – 35 % on favourites
33.7 %won1.08fair A/E120,749N
The favourite, ridden by a rider who wins 36 % or more on favourites won 38.4 % of 39,618 runs at fair A/E 1.10; the favourite, ridden by a rider under 30 % on favourites 30.9 % of 46,349 at 1.09. Verdict: true — and priced.
Favourites, by their rider's record on earlier favourites as it stood the day before. The rows differ on strike by construction; the fair A/E says whether the market prices the rider.
The last 365 days: the favourite, ridden by a rider who wins 36 % or more on favourites 39.2 %, fair A/E 1.06 (N 2,837) · the favourite, ridden by a rider under 30 % on favourites 30.8 %, fair A/E 1.04 (N 2,489)
Since we first published this, on 28 October 2026: the favourite, ridden by a rider who wins 36 % or more on favourites: too few to say (N 17) · the favourite, ridden by a rider under 30 % on favourites: too few to say (N 11)
Every GB and Irish race from 2 January 2005 to 29 October 2026; each record read as it stood the day before the race. N is runs; fair A/E over runs with a usable starting price.
Jockey statistics →
“Some riders just go well at a track.”
True — and pricedAdded 28 OctoberVerdict: Some riders do go well at a track: a rider with a strong record at a course wins more often there than one with a poor one, and the market prices it. Each course's page lists them. The price already says so: on this alone the market gives nothing away.
- A rider 4 points or more better here than overall
14.7 %won1.03fair A/E148,002N
- A rider 4 points or more worse here than overall
10.5 %won1.01fair A/E140,556N
- A rider within 4 points of their overall here
10.9 %won1.01fair A/E1,138,713N
A rider 4 points or more better here than overall won 14.7 % of 148,002 runs at fair A/E 1.03; a rider 4 points or more worse here than overall 10.5 % of 140,556 at 1.01. Verdict: true — and priced.
The rider's record at the course against their record everywhere, as both stood the day before. Each course's page lists the riders who go well there.
The last 365 days: a rider 4 points or more better here than overall 15.6 %, fair A/E 1.01 (N 7,186) · a rider 4 points or more worse here than overall 11.0 %, fair A/E 1.03 (N 7,001)
Since we first published this, on 28 October 2026: a rider 4 points or more better here than overall: too few to say (N 49) · a rider 4 points or more worse here than overall: too few to say (N 19)
Every GB and Irish race from 2 January 2005 to 29 October 2026; each record read as it stood the day before the race. N is runs; fair A/E over runs with a usable starting price.
Racecourse statistics →
“Follow the first-season sires.”
Not in the numbersAdded 28 OctoberVerdict: The first-season sire is a sales story. Its early runners win at the rate their price says, no more than an established sire's, so there is nothing in following the new name.
- A first-season sire's runner (first three career runs)
9.8 %won1.00fair A/E35,167N
- An established sire's runner (first three career runs)
9.9 %won0.99fair A/E155,569N
- A second-season sire's runner (first three career runs)
9.4 %won1.00fair A/E38,261N
A first-season sire's runner (first three career runs) won 9.8 % of 35,167 runs at fair A/E 1.00; an established sire's runner (first three career runs) 9.9 % of 155,569 at 0.99. Verdict: not in the numbers.
Flat runners in their first three career runs, by how long their sire has had runners. The first-season sire is the one the sales talk is about.
The last 365 days: a first-season sire's runner (first three career runs) fair A/E 0.94, won 9.7 % (N 1,344) · an established sire's runner (first three career runs) fair A/E 0.98, won 10.7 % (N 11,599)
Since we first published this, on 28 October 2026: a first-season sire's runner (first three career runs): too few to say (N 9) · an established sire's runner (first three career runs): too few to say (N 99)
Every GB and Irish race from 5 January 2007 to 29 October 2026; each record read as it stood the day before the race. N is runs; fair A/E over runs with a usable starting price.
Horse statistics →
Questions
Do non-runners make the favourite a better bet?
Not in the numbers. The missing runners are already in the price. With three or more gone the favourite wins no more often than its new odds say — if anything a little less — so there is nothing extra in the jolly once the race has fallen apart. The favourite in a race that lost 3 or more declared runners won 29.3 % of 5,012 runs at fair A/E 1.08; the favourite in a race that lost none 32.6 % of 267,741 at 1.09. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Do favourites win more often early in the day?
True — and priced. The early favourite does win more often — but that is the race mix, small fields and the all-weather before lunch, and the price knows it. There is no safety in the hour itself. The price already says so: on this alone the market gives nothing away. The favourite, off before 1 pm won 37.9 % of 8,840 runs at fair A/E 1.11; the favourite, off from 5 pm 31.5 % of 104,283 at 1.09. Verdict: true — and priced. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Is a yard's second string a good bet?
Not in the numbers. The dark horse is a myth. The yard's longer-priced runner wins far less often than its stablemate and, worse, runs below even its longer price: the market still gives it too much credit for the yard's name. The longer-priced stablemate won 4.0 % of 151,865 runs at fair A/E 0.87; the shorter-priced stablemate 11.7 % of 150,951 at 1.01. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Are odds-on favourites bad value?
Not in the numbers. Odds-on shots are the opposite of bad value in fair terms. They win more often than their price implies — the favourite–longshot bias at its purest — so the complaint is about the small return, not the price. Flat: Odds-on (fair) runners on the flat won 63.5 % of 12,113 runs at fair A/E 1.08 against the fair line of 1.00. Verdict: not in the numbers. Jumps: Odds-on (fair) runners over jumps won 65.1 % of 12,402 runs at fair A/E 1.09 against the fair line of 1.00. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Is heavy going a lottery?
Not in the numbers. Heavy going is not a lottery in either code: the favourite wins at least as often as on good, and no more races go to a big price. Jumps: The jumps favourite on heavy won 36.2 % of 15,891 runs at fair A/E 1.12; the jumps favourite on good 33.8 % of 39,869 at 1.08. Verdict: not in the numbers. Flat: The flat favourite on heavy won 28.9 % of 5,510 runs at fair A/E 1.03; the flat favourite on good 30.2 % of 34,223 at 1.07. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
On heavy going, are only the short-priced favourites worth backing?
Not in the numbers. Backwards. On heavy it is the bigger-priced favourite that runs above its price, not the odds-on one, which behaves on heavy exactly as it does everywhere else. The favourite at 3/1 or longer (fair) on heavy won 21.6 % of 8,332 runs at fair A/E 1.14; the odds-on favourite on heavy 64.8 % of 2,300 at 1.09. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Is racing harder to read around the autumn changeover?
Not in the numbers. The favourite in the autumn changeover window wins as often as in summer in both codes: the changeover is a story, not a season. Flat: The flat favourite, 15 October – 15 November won 29.7 % of 12,714 runs at fair A/E 1.09; the flat favourite, May – September 31.6 % of 108,379 at 1.09. Verdict: not in the numbers. Jumps: The jumps favourite, 15 October – 15 November won 34.9 % of 13,024 runs at fair A/E 1.10; the jumps favourite, May – September 33.0 % of 36,034 at 1.09. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Does a jumps favourite need the run after a summer break?
Not in the numbers. "Needs the run" is not in the numbers. The autumn jumps favourite off a summer break wins as often as one that raced within the last two months — the yards that send a fresh favourite out send it fit. The autumn jumps favourite off 150 days or more won 36.8 % of 4,465 runs at fair A/E 1.10; the autumn jumps favourite raced within 60 days 34.1 % of 7,286 at 1.11. Verdict: not in the numbers. Every GB and Irish race from 15 October 2005 to 29 October 2026.
Is the spring changeover easier to read than the autumn one?
Not in the numbers. Not the easier one. The flat favourite in the spring window wins no more often than in the autumn, and the market reads it worse there than in any other season. The flat favourite, 15 March – 30 April won 30.8 % of 17,979 runs at fair A/E 1.05; the flat favourite, 15 October – 15 November 29.7 % of 12,714 at 1.09. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Do three-year-olds beat older horses in the autumn?
Half true. It depends on the month. The weight-for-age scale is designed to level the generations and in most months the market prices it to the pound; where the three-year-olds run above their price it is one month, not the autumn. September: Three-year-olds against older horses, September won 10.5 % of 38,273 runs at fair A/E 1.02; older horses against three-year-olds, September 8.3 % of 64,226 at 0.99. Verdict: not in the numbers. October: Three-year-olds against older horses, October won 9.6 % of 36,430 runs at fair A/E 1.03; older horses against three-year-olds, October 7.8 % of 59,506 at 0.98. Verdict: true. November: Three-year-olds against older horses, November won 9.9 % of 17,528 runs at fair A/E 1.02; older horses against three-year-olds, November 8.5 % of 33,077 at 0.99. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Can the top weight win a big-field handicap?
Not in the numbers. The top weight is the horse to fear, not to dismiss. In a big-field handicap it wins more often than the bottom weight, not less, and is fairly priced doing it — it carries the most because it is rated the best, and the mark wins more than the weight costs. The top weight in a handicap of 14 or more won 8.3 % of 35,151 runs at fair A/E 1.03; the bottom weight in a handicap of 14 or more 4.3 % of 34,038 at 0.97. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Does weight tell more on soft ground?
Half true. Half true: weight tells in the mud on the flat only. The flat handicap top weight runs below its price on soft or heavy and above it on faster ground; over jumps the top weight runs to its price whatever the going. Flat: The flat handicap top weight on soft or heavy won 12.4 % of 15,125 runs at fair A/E 0.97; the flat handicap top weight on good or faster 14.8 % of 103,365 at 1.03. Verdict: true. Jumps: The jumps handicap top weight on soft or heavy won 13.4 % of 24,033 runs at fair A/E 1.00; the jumps handicap top weight on good or faster 14.2 % of 29,383 at 1.00. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Is a 7 lb claiming rider worth the allowance?
Not in the numbers. The allowance is priced to the pound. A horse with a claimer up wins at the rate its price says — the bigger the claim, if anything, the less — so the pounds off are not a free gift. Ridden by a 7 lb or heavier claimer won 7.0 % of 271,818 runs at fair A/E 0.97; no claim 11.0 % of 1,895,412 at 1.01. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Are horses aged ten and over worth backing?
Not in the numbers. The old boys are over-backed. Horses aged ten and over run below their price — the market pays for the reputation, and the reputation is worth a little less than it costs. Aged ten and over won 7.6 % of 117,950 runs at fair A/E 0.94 against the fair line of 1.00. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Do fillies and mares lose out against males on the flat?
Not in the numbers. Fillies and mares hold their own against males on the flat: they win less often but the allowance covers it closely enough that the market gap is within our threshold. The belief is not in the numbers. Fillies and mares against males won 8.5 % of 382,637 runs at fair A/E 0.98; males against fillies and mares 10.4 % of 837,216 at 1.01. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Is a horse back on the track within three days being rushed?
Not in the numbers. The opposite of rushed. A horse back out within three days runs above its price in both codes — a yard brings a horse straight back because it came out of the last race well, and the market under-reads that. Jumps: Jumps, back within 3 days won 14.1 % of 5,670 runs at fair A/E 1.15; jumps, 8 – 21 days since the last run 10.1 % of 307,181 at 0.98. Verdict: not in the numbers. Flat: Flat, back within 3 days won 14.4 % of 19,658 runs at fair A/E 1.09; flat, 8 – 21 days since the last run 10.6 % of 637,689 at 0.99. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Does first-time headgear improve a horse's chance?
Not in the numbers. Blinkers are not a magic switch. First-time headgear runs no better than its price in either code; the change is already in the odds, and in the record it is slightly cold. Flat: First-time headgear, flat won 8.2 % of 63,966 runs at fair A/E 0.95; no first-time headgear, flat 10.3 % of 1,497,766 at 1.00. Verdict: not in the numbers. Jumps: First-time headgear, jumps won 8.0 % of 44,646 runs at fair A/E 0.98; no first-time headgear, jumps 10.3 % of 1,007,839 at 1.00. Verdict: not in the numbers. Every GB and Irish race from 3 January 2005 to 29 October 2026.
Do course-and-distance winners win more often?
True — and priced. Horses for courses holds on the strike rate: the course-and-distance winner wins more often than a horse on its first visit. The price already says so: on this alone the market gives nothing away. A course-and-distance winner won 13.0 % of 269,365 runs at fair A/E 1.00; first run at the course 10.0 % of 1,486,947 at 1.00. Verdict: true — and priced. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Does a horse's first run for a new yard improve its chance?
Not in the numbers. A change of yard works no oracle. The first run for a new yard is, in the record, a run below the price: the market gives the move more credit than it earns. First run for a new yard won 8.4 % of 102,521 runs at fair A/E 1.02; same yard as the last run 10.5 % of 2,348,840 at 1.00. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Is a horse dropping in class a good bet?
Not in the numbers. "Thrown in" is already in the price. A horse down in class wins more often, as it should, but runs to its odds, not above them: the drop is the first thing the market sees. Down in class from the last run won 12.1 % of 416,873 runs at fair A/E 1.01; same class as the last run 11.1 % of 1,012,335 at 1.00. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Is a horse stepping up two furlongs or more a good bet?
Not in the numbers. The step up in trip is not an edge. A horse going two furlongs or more further than last time runs to its price, not above it: whatever the trip will bring out, the market has allowed for. Two furlongs or more further than the last run won 8.8 % of 314,383 runs at fair A/E 0.99; the same distance as the last run 11.4 % of 778,782 at 1.01. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Do races on official 'Good' ground beat the course standard time?
Not in the numbers. "Good" is slow on the clock. Only a small share of flat races on official good ground finish inside the course standard — standards are near-record times, so running inside one is rare on any going. The useful reading is how far outside it a course usually runs, which is on each course's page. Of 29,317 flat races run on official 'Good' going, 3.2 % finished inside the course standard time; on average the winner was 0.46 seconds per furlong outside it. Every GB and Irish race from 19 March 2005 to 4 November 2025.
Do horses that won their last two runs keep winning?
True — and the market has not caught up. Winners do keep winning. A horse that won its last two wins far more often than one that did not win last time, and the hat-trick bid is one of the few streaks the market under-prices. And the market has not caught up: this group has been running above its price. Won its last two runs won 22.0 % of 46,637 runs at fair A/E 1.04; did not win last time 9.2 % of 2,399,668 at 0.99. Verdict: true — and the market has not caught up. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Does a handicapper need to be back on its last winning mark to win?
Not in the numbers. Backwards. A handicapper running above its last winning mark wins more often than one that has dropped below it, not less: being above the mark is the sign of a progressive horse the handicapper is chasing, not a bar. Running 5 lb or more above its best winning mark won 12.1 % of 306,914 runs at fair A/E 1.02; running 4 lb or more below it 8.9 % of 246,579 at 0.99. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Can a horse beaten at odds-on last time be trusted next time?
Not in the numbers. The opposite. A horse beaten at odds-on last time bounces back next time more often than one beaten at a longer price, and runs above its new odds: punters desert it and the record says they should not. Beaten at odds-on (fair) last time won 25.1 % of 8,134 runs at fair A/E 1.09; beaten at 3/1 or shorter (fair) last time 19.0 % of 112,629 at 1.04. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Does a well-beaten debutant improve for the run?
Not in the numbers. A well-beaten debutant does not improve for the run — or not as much as the market hopes. Seventh or worse on debut is the worst second-run bet in the record, running well below its price. Seventh or worse on debut won 3.5 % of 85,627 runs at fair A/E 0.87; fourth to sixth on debut 9.9 % of 53,235 at 0.97. Verdict: not in the numbers. Every GB and Irish race from 3 January 2005 to 29 October 2026.
Is a horse beaten twenty lengths or more last time overpriced next time?
Not in the numbers. Not overpriced. A horse beaten twenty lengths or more last time still runs below its price next time: the market forgives the heavy defeat a little too readily. Beaten 20 lengths or more last time won 5.2 % of 96,301 runs at fair A/E 0.94; beaten half a length or less last time 10.9 % of 517,415 at 1.01. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Is an unexposed horse running in its first handicap a good bet?
Not in the numbers. "Unexposed" is priced. The lightly-raced horse in its first handicap wins more often than the handicap regular, as the market expects, and runs to its odds: there is nothing left over for knowing it is unexposed. First handicap after one or two runs won 13.6 % of 12,880 runs at fair A/E 1.01; has run in a handicap before 10.0 % of 1,452,314 at 1.00. Verdict: not in the numbers. Every GB and Irish race from 3 January 2005 to 29 October 2026.
Is a horse's own record on today's going the key?
True — and priced. It wins more often on ground it has won on before — but the market knows the form book too, so the horse's own going record is already in its price. The ground signal that is not priced is the sire's, on the horse's page. The price already says so: on this alone the market gives nothing away. Has won on today's going before won 12.2 % of 778,019 runs at fair A/E 1.01; first run on today's going 9.2 % of 728,972 at 0.99. Verdict: true — and priced. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Does a returning horse's chance depend on its yard's record with fresh horses?
True — and the market has not caught up. It depends on the yard. A horse off six months from a yard that wins with its fresh ones wins several times as often as one from a yard that does not, and the market has not caught up: it prices the days off, not the yard's record with them. And the market has not caught up: this group has been running above its price. Off 180+ days, from a yard that wins 15 % or more with fresh horses won 19.0 % of 34,207 runs at fair A/E 1.09; off 180+ days, from a yard under 8 % with fresh horses 5.2 % of 66,322 at 0.90. Verdict: true — and the market has not caught up. Every GB and Irish race from 7 July 2005 to 29 October 2026.
Is a top rider booked by a quiet yard a sign?
Not in the numbers. The booking on its own means nothing. A top rider for a quiet yard's horse runs exactly to its price: the market reads the booking too, and the horse is still the quiet yard's horse. The rider adds to a hot yard, not to a cold one. A top rider (18 % or more over 90 days) for a quiet yard (under 8 %) won 11.9 % of 25,787 runs at fair A/E 1.00 against the fair line of 1.00. Verdict: not in the numbers. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Does the rider's record on favourites change the favourite's chance?
True — and priced. A favourite ridden by a rider with a strong record on favourites does win more often — and the market has priced the rider in: both favourites run to their odds, so the top jockey is not a reason on his own. The price already says so: on this alone the market gives nothing away. The favourite, ridden by a rider who wins 36 % or more on favourites won 38.4 % of 39,618 runs at fair A/E 1.10; the favourite, ridden by a rider under 30 % on favourites 30.9 % of 46,349 at 1.09. Verdict: true — and priced. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Do riders with a good record at a course win more there?
True — and priced. Some riders do go well at a track: a rider with a strong record at a course wins more often there than one with a poor one, and the market prices it. Each course's page lists them. The price already says so: on this alone the market gives nothing away. A rider 4 points or more better here than overall won 14.7 % of 148,002 runs at fair A/E 1.03; a rider 4 points or more worse here than overall 10.5 % of 140,556 at 1.01. Verdict: true — and priced. Every GB and Irish race from 2 January 2005 to 29 October 2026.
Are a first-season sire's early runners worth following?
Not in the numbers. The first-season sire is a sales story. Its early runners win at the rate their price says, no more than an established sire's, so there is nothing in following the new name. A first-season sire's runner (first three career runs) won 9.8 % of 35,167 runs at fair A/E 1.00; an established sire's runner (first three career runs) 9.9 % of 155,569 at 0.99. Verdict: not in the numbers. Every GB and Irish race from 5 January 2007 to 29 October 2026.
What does fair A/E mean on this page?
Fair A/E is winners divided by the winning chances implied by the starting prices once the bookmakers' margin is removed. A row above 1.00 won more often than its price said; below 1.00, less often. Favourites tend to sit above 1.00 and long shots below it (the favourite–longshot bias), so compare rows with rows, not with 1.00 alone.
How is each verdict decided?
By a rule, not by a writer. A belief about who wins more often is "true" when the group it names wins at least two points of strike more than the comparison group, and "priced" when their fair A/E are within 0.04 of each other. A belief about who is the better bet is "true" when the fair A/E gap is at least 0.04 in the direction it predicts. Otherwise the belief is "not in the numbers". A cell with too few runs says so instead of printing a number.
How often are these numbers updated?
Nightly. Every claim is summed from a store rebuilt from our archive after each day's results, so each number carries the date of the last race it includes. The day a claim first appeared on this page is recorded, and from then on it also shows what has happened since.