Explaining my ratings for the AW what they mean

[tony spencer]

This is what the finishing profile means

 

[tony spencer]

This is what the finishing profile means this is the ratings and profiles. These are all based on sectional times The finishing edge.

 

[tony spencer]

Todays ratings Chelmsford

 

[tony spencer]

todays bets wolve

 

[tony spencer]

Early Pace Prediction Model – Overview Pace Graphs Part​

I’ve built a machine learning model that predicts how a horse race will unfold early on, using historical race data.


The idea is to estimate which horses are most likely to lead or race prominently in the early stages, because early pace is one of the biggest factors influencing how a race develops.


The model works in two stages.

---

Stage 1 – Predicting Early Pace for Each Horse​

For every runner in a race the model analyses:

• The horse’s last 10 runs
• Historical early pace rankings
• Previous pace ratings
• The track they are racing at
• Various statistical summaries of their past performances

Using this information the model estimates the probability that each horse will:

1. Lead the race
2. Race in the front 2 early
3. Race in the front 3 early

This is done using gradient boosting machine learning models trained on several years of race data.

---

Model Accuracy​

On unseen test data the model performs as follows:

Prediction	Top Ranked Horse Accuracy
Predicting the leader	35% correct
Predicting a horse in the first 2 early	55% correct
Predicting a horse in the first 3 early	67% correct

What this means​

If you simply take the horse the model ranks #1:

• It correctly identifies the actual early leader about 35% of the time.
• The top-ranked horse is in the first two early positions about 55% of the time.
• The top-ranked horse is in the first three early positions about 67% of the time.

Given that most races have 8–12 runners, these hit rates are significantly better than random selection.

---

Stage 2 – Predicting Overall Race Pace​

Once the model estimates how each horse is likely to run early, it then analyses the whole field together.


From this it predicts whether the race is likely to be:

• Very Slow
• Slow
• Standard
• Fast
• Very Fast

This helps understand how the race is likely to be run tactically, which is very useful when evaluating horses that prefer strong pace or slower races.

---

Why This Matters​

Early pace often determines:

• Which horses get the best tactical position
• Whether closers or front runners are favoured
• How strongly the race is run overall

By modelling pace before the race, it becomes possible to anticipate the likely race shape rather than just analysing past form.