Four days on the Knavesmire begin with the race most years crowned the best run in Britain, and this year the Juddmonte has earned the billing before a stall opens: the defending champion, a Royal Ascot winner rated the highest in training, against an unbeaten three-year-old who took the Eclipse last month. Around it sits everything that makes York in August the deepest week of the season, two cavalry-charge Heritage Handicaps, a stayers' marathon with no pace to run to, a juvenile Group 3 with a hot stable double, and a closing nursery that hides a Goodwood rematch regular readers will remember.
This guide reads all seven races through our model, and it works the way our coverage always works: the framing of every race is open to everyone, a free account unlocks the complete analysis on race morning, and members get the live data tables underneath, the same percentages, ratings and pace maps the dashboard runs, updating through the day. Along the way we explain the figures as we use them, because the point of these guides has never been tips. It is showing how data narrows twenty-two runners to two or three worth your attention, and letting you take the method anywhere.
New to our work? Two minutes here and every number below will make sense. Tap any question to open it.
TPR (Total Performance Rating) is EquiAnalytix's proprietary rating, built from a database of more than 836,000 runs. Every horse's number is derived from race times adjusted for class, going, pace and race shape, so it measures what a horse actually did rather than what the handicapper thinks of it.
Every horse in a race holds a TPR percentage and every field sums to 100, so the numbers read like the model's own market. Across our whole database, 73 percent of winners come from our top four ranked horses, which is why the top of the list is where every read begins.
Ability is the ceiling: the underlying class a horse has shown at its very best. Operating is the now: the level it is producing this season, Ability plus current form. When both agree at a high level, as with the best horses on this card, the model reads a horse at its peak.
Going, track and distance figures isolate a horse's best performances under today's exact conditions. A high going figure on Good ground matters far more than a career best earned in a bog, and a big track figure says this course specifically brings out the horse's best. They also remember runs that finishing positions forget, as with beaten horses who met track bias.
Each runner's early-position style is profiled from its history, and the race pace forecast adds them up. High numbers mean contested speed and favour those who can sit just off it; very low numbers mean a crawl and favour the prominent. At York specifically, two decades of data say front-runners outperform expectation, especially in sprints.
The gap between the official mark a horse carries today and the mark it last won from. Small numbers mean the handicapper has not yet punished the horse for winning; big numbers mean it must be better than ever to win again. Across our database, handicaps disproportionately go to the small end of the scale.
Zeros mean absence of evidence, not a verdict. A horse whose form is entirely overseas, like a French raider, sits outside our database, and we would rather tell you the model cannot see it than dress a blank up as an assessment.
Want the full walkthrough of every metric on the platform, with examples? Read the complete EquiAnalytix data guide →
Twenty-two sprinters, a Heritage Handicap pot, and a genuine collision between what the race looks like and what this course usually does with it. Our pace forecast for the opener is 11, among the highest numbers we have published for any race this season: a wall of confirmed early speed that ought, in theory, to burn itself out and hand the finish to the closers. York's own history argues back. Over nearly two thousand sprint runs on the Knavesmire, horses racing on the lead have outperformed expectation by 36 percent, while the hold-up runners have badly underperformed. When the model's percentages also spread as thinly as they do here, with just four points covering the top four, the data is telling you two things at once: keep your reading of the race wide, and think hard about who controls the front.
The full analysis of every race is free to read
One free account unlocks every race's write-up here and in all our guides, all season. No card, ten seconds.
Unlock every race freeAlready have one? Sign in
Eleven two-year-olds, thirty-one career runs between them, and a Group 3 that has launched more than its share of subsequent stars. Races like this are where our data works differently: with so little form on the page, the model leans on the things that project rather than the things that are proven, the sire's record with juveniles, the shape of a stable's season, the size of the figures earned in a handful of starts. One yard in form saddles two of the principals, the market's likely favourite has to fight another confirmed speedball for the early lead, and the biggest last-time figure in the race belongs to neither of them.
Six runners, and the smallest field of the day carries the deepest form: the winner and the runner-up from Epsom's Derby meet again, with the Irish Derby third in between them on the racecard and a third Ballydoyle runner whose likely role is to make sure the gallop suits the other two. When the model puts a 42 on one horse, as it does here, it is about as close to conviction as our numbers ever get. The interest is less in whether the percentages are right than in what price the rematch angle offers against them.
Most years the Juddmonte is billed as the best race run in Britain, and most years the billing struggles to live up to itself. Not tomorrow. The horse who won this race twelve months ago comes back as a Royal Ascot winner with the highest official rating in training, and the thing sent to beat him is an unbeaten three-year-old who took the Eclipse last month and meets his elder on weight-for-age terms. Between them on our numbers sits a Group 1 winner whose two best career figures were both earned on this track, and lurking outside the numbers entirely is a French raider our database, honestly, cannot see. The model has made its call, and made it emphatically. Championship days are why the data exists.
Two miles, seventeen runners, and a pace forecast of just 2, the lowest of the day. Stamina races with no pace are a lovely paradox: the trip says endurance, the tempo says sprint finish, and the result usually belongs to whoever is placed when the dash begins. Our model's top pick arrives with a complication the percentages keep quiet about, one in-form yard runs a horse for each version of the race, and the in-form progressive in the field is stepping into the unknown on trip. A proper puzzle, and the shape of it matters more than any single number.
Nineteen sprinters and a pace forecast of 16, the highest number our model has ever attached to a race in these guides. If the opener asks what happens when too much speed meets York's five furlongs, this one asks it at full volume: seven of the first nine on our list are confirmed front-runners, and the percentages spread just seventeen, ten, seven across the top three. When the model declines to commit, that is not the data failing, it is the data being honest, and the reading below is built accordingly: a race to attack with a wide net and a clear head about which speed survives.
The day closes with a nursery, a handicap for two-year-olds, which is the purest contest on any card between the handicapper's guesswork and everybody else's. With babies, the official assessor is rating a handful of runs and a lot of projection. So are we, but our projection carries figures, and in a field of twenty-two they point two very different stories at the top: a maiden the clock adores, and a filly who has stopped being beaten. There is also a Goodwood rematch buried in here that regular readers of our festival coverage will remember.
18+ | Please gamble responsibly | GamCare 0808 8020 133. All figures are EquiAnalytix proprietary ratings derived from our 836,398-run database; they are analysis, not betting advice.