Welcome to our full model predictions.

As a member on a full model tier, you will have access to our model's race outcome probabilities. These are scheduled for every race day, for all rules races.

What to expect

The pages in this section contain live-updated tables of predictions for each race. Each horse is given a probability of winning and of being placed, and is ranked by the win probability. Further details below:

Race types

All rules races are covered: flat and jumps ("national hunt") races, excluding things like point-to-point races.

Country

British and Irish race courses.

Timing

Predictions are first published one day in advance of the race, in the evening (usually between 8 and 9pm UK time).

Live updates

The tables are updated in response to new information such as non-runner announcements and going updates. Either of these factors could change the race outcome probabilities. So, if there is a race you are particularly interested in, it is worth refreshing the page as race time approaches.

If not all races for the same day appear at the same time, just wait - the remainder should show up within about an hour. Sometimes predictions for either one of flat or jumps races will be published before the other; sometimes races for certain courses will appear before others.

Explanation of the model output

The table above is an example of the model output (from the Aintree Festival 2025). At the top is the race title. The course, date and time will also appear on the page. Here is an explanation of each data point in the table:

Number: The "cloth number"; the number the horse wears, which you will also see on the racecard or a bookmarker's board.

Name: The name of the horse.

Win chance %: Probability of winning, to the nearest percent. These will add up to 100% (except for rounding errors). Table rows are ordered in descending order of win chance, by default (you can re-order the live tables by clicking column headers).

Place chance %: Probability of being placed (an outcome which depends on the number of runners and whether the race is a handicap), to the nearest percent. When there are fewer than 5 runners, this is the same as the win chance.

Relative pros/cons: Data points which the model thinks count in favour of this horse or against this horse. Points in favour are given plus (+) signs, points against are given minus (-) signs. Note these are relative to other horses in the race. See the section below for more details.

Non-runner?: Has the horse been withdrawn? 1 means yes, 0 means no.

Model odds: The odds against this horse winning, in standard format. This is derived from the win chance. It represents the minimum odds that should be considered good value for this runner, according to the model. In brackets are the minimum odds that should be considered good value for an each-way bet (only shown if there are at least 5 runners). Each-way bets sometimes require longer (or shorter) odds to be considered worthwhile.

More details on the "relative pros/cons" column

This column gives an insight into which factors are tending to move a horse's win probability up or down. Besides giving some explanation of how the model has arrived at its predictions, these pointers could be useful in their own right for race analysis.

The possible data points reported are:

• Quality
• Trend
• Going
• Distance
• Jockey
• Trainer
• Sire
• Dam
• Draw

Quality measures the strength of a horse in average conditions; its tendency to beat other horses. Trend measures whether performance is improving or degrading over time. The remaining factors are self-explanatory. Note that draw (stall) will only appear for flat races.

As noted above, the pros/cons are relative to other horses in the race. So, for example, "-jockey" does not mean this runner's jockey is bad necessarily, just that relative to the other jockeys in the race, this jockey is adversely impacting the win chances of this runner, according to the model.

Some runners may have more than one pro/con, some runners may have none at all. Some pros/cons may be even stronger pointers; these are given additional plus/minus signs.