Saturday is the centrepiece of Cheltenham's November Meeting: fifteen runners in the Paddy Power Gold Cup, the Grade 2 Triumph Trial and Arkle Trial before it, three more handicaps around it and a Listed mares' bumper to finish, all on Good ground.
This guide reads all seven through our model. It is written for anyone who has never seen our numbers before: every figure is explained the first time it appears, and each race shows a different way of working with the data, from a clear leader in the opener to a bumper where most of the field has no form to measure. The framing of every race is open to everyone, a free account unlocks the complete analysis, and members get the live racecard underneath. The third race, the Arkle Trial, is also our free race of the day, open in full on the website.
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 machine-learning rating. It turns everything we hold on a horse into its percentage chance of winning. It was tested on 836,398 runners rated before their races across GB and Irish racing from 2019 to 2026: the top-rated horse won 28.8% of the time and the top four included 72.7% of winners.
Our rating engine re-reads every horse's career after every run. Ability is the level a horse has proved it can run to. Oper, short for Operating, is the level it is ready to run to today. Both sit on the same scale as the official handicap mark, in pounds.
The gap between Oper and Ability, in pounds. A plus figure means a horse is running above its own proven level; a minus figure means it is running below it.
The gap between our Ability figure and the horse's official handicap mark. A plus figure means our numbers rate the horse higher than the handicapper does.
The engine needs runs to measure. A horse that has never run under Rules, or whose form is not in our database, shows a blank beside Ability and Oper. Its percentage is then built from breeding, stable form and jockey.
Every run earns a figure built from the time of the race and adjusted for its class, so runs at different tracks and in different grades can be compared. An easy win in a small field can earn a low figure, because the horse was never asked to run fast.
Yes. French races sit in our database alongside British and Irish ones, with performance figures on the same scale, so a horse's French runs can be read next to its British ones.
The level a horse has run to on ground like today's, at today's track and at around today's trip. A blank means it has never raced under those conditions.
The spread of a horse's recent figures. A low number means it runs to a similar level every time.
The sire type figure is the average performance figure of a sire's runners in today's kind of race, such as chases or bumpers. The dam form figure is the level the mother herself ran to over jumps.
We read each runner's usual early position from its recent races and score the likely gallop from 0 (no pace at all) to 9 (very fast).
A yard's strike rate, winners divided by runners, over the last month, shown beside its figure for the year, along with the average performance figure of its runners.
Want the full walkthrough of every metric on the platform, with examples? Read the complete EquiAnalytix data guide →
Twelve runners over three miles and a furlong, every one of them ridden by an amateur, and our model has a clear leader. TPR, our Total Performance Rating, turns everything we hold on a horse into its percentage chance of winning, and LITTLE LEDGEND's is 23. Nothing else is above 11: LADIES WELL 11, TOP GUY 10, JACCOURS 10, DINOLAND 9, KARNAVAL POINT 8. Tested on 836,398 runners rated before their races, the top-rated horse on TPR has won 28.8 per cent of the time, so a clear leader is a strong starting point and no more than that. The rest of the racecard is there to tell you why he leads and what could beat him.
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
A different kind of puzzle. These are nine three-year-olds at the start of their hurdling careers, most with one or two runs over hurdles behind them, so there is far less form to read than in the opener. TPR reflects that by keeping them close: ATOMIC CITY 21, CHARLES DARNAY 19, MAGNETUDE 16, MIX MATCH 13, ADMIRAL WILL BROWN 11, DON DANDY 11. When the evidence on the track is this thin, the model leans harder on what a horse did on the Flat, who trains it and how it is bred, and the racecard lets you see each of those for yourself.
Seven go to post for the Arkle Trial and our model gives one of them almost half the race. KABRAL DU MATHAN is on 49, the biggest percentage on the card, with OLD PARK STAR 22, WORKAHEAD 11, NEWTOWN RAMBLER 7 and PASSING DIAMOND 6. It is worth reading that number properly. A 49 says he is far more likely to win than anything else. It also says that, slightly more often than not, something beats him. That is what an honest percentage looks like in a seven-runner novices' chase. This is also our free race of the day, so the full racecard, with every number below and many more, is open on our website for anyone to explore.
The feature, and the race where a single percentage tells you least. Fifteen runners, and the highest TPR in the field is 14: BLUEKING DOROUX 14, ZURICH 11, TORNEO 10, then QUEBECOIS, JULY FLOWER and NORN IRON on 8 each. In a handicap this open the more useful statistic is a wider one. Across those 836,398 runners, 72.7 per cent of winners came from TPR's top four, so the model's job here is to shrink fifteen to a shortlist, and the rest of the racecard is how you choose within it.
Sixteen runners over three miles, and our model cannot split the top two: ACE OF SPADES 14, AIMERIC 14, WILLIAM YEATS 12, GOLD FOR ALEC 11, with nothing else above 7. When TPR is level like this, the engine is the tie-break we reach for. Members can sort any racecard by Ability or by Oper, and in this race the two lists are headed by two different horses.
Two horses share top spot on the model and they could hardly be more different. DAYLATEDOLLARSHORT and MONEYGARROW are both on 19, with GRAND CONQUEROR 13, SUNNY SOUTH WEST 9 and A KING OF MAGIC and SEAMOON STORY on 7. One of the pair ran five weeks ago and won. The other has not been seen since the Festival in March and has never run in a handicap. It is hard to look past the top two, so the useful question is what separates them.
The card ends with the race that tests a data model hardest. A bumper is a Flat race for jumping-bred horses who have barely started, and nine of these fifteen mares have never given our engine a run to measure. So this is the place to be straight about what a percentage means when the form book is nearly empty. TPR still gives every runner a chance, SOLERETTA 16, NORTHERN SINGER 13, CLOVELLY CHARM 12, then five mares on 9, but here it is built mostly from breeding, stable form and jockey, because there is little else. The spread is narrow for that reason.
Every big raceday we read the whole card with the model: ratings, speed figures, pace and draw. We send the next one to your inbox the morning it is published. No charge, and you can unsubscribe in one click.
We use your email to send racing guides and occasional news about EquiAnalytix. We never share it. 18+ | Please gamble responsibly.
18+ | Please gamble responsibly | GamCare 0808 8020 133. All figures are EquiAnalytix proprietary ratings derived from our database of every UK and Irish race since 2009; they are analysis, not betting advice.