TrapMetricsRatings and probabilities: two different questions
What a rating says about a runner, what a probability says about uncertainty, and why a 30% chance still leaves 70% for the rest of the field.
A comparison and an estimate
A rating and a probability help with different questions. A rating gives you a consistent way to compare runners. A probability tries to express how uncertain the outcome is. Neither removes the need to understand the field.
On TrapMetrics the two are tied together on purpose. Every runner gets a win chance first, and the TrapMetrics Rating (TMR) is that chance put on a handicap scale: TMR 100 is an average chance in that field, and every 15 points doubles it. So TMR 115 means twice the average rival's chance, and 85 means half.
What does 20% actually mean?
Imagine many comparable predictions that each give a runner a 20% chance. If those estimates are well calibrated, roughly one in five of those runners should win. That is a long-run relationship across predictions. It does not say which individual runner will win.
It also does not mean every block of five produces exactly one winner. Outcomes vary. That variation is one reason a handful of examples cannot validate a model, and why the evidence page scores calibration by band on thousands of settled races rather than on the winners people remember.
The other runners still matter
Read probabilities as a view of the whole field. In a race with one winner and every runner accounted for, the win chances add to 100%. If one runner is estimated at 30%, the other runners together account for 70%.
A real example: at Nottingham at 13:21 on 20 August, Kellies Jewel was the top-rated dog on our card with a 30.9% chance, and won at a Betfair SP of 6.0. Before the off, the other five dogs still held more than two thirds of the chance between them. "Most likely runner" and "more likely to win than not" are different statements, and the second is rarely true in a six-dog race.
Use the number to choose the next question
Open the runner's form. Look at the track context. Check the method and the evidence. A useful racecard makes that investigation easier rather than turning an estimate into a command.
The market is part of the same picture. The closing Betfair price beats every rating over time, ours included, so a chance on our card is most useful where it differs from the price and you can see why.
Questions this piece answers
What is the difference between a rating and a probability?
A rating gives you a consistent way to compare runners with each other. A probability estimates how likely each outcome is, so the chances across a field add up to 100%. On TrapMetrics the rating is derived from the probability: TMR 100 is an average chance in that field, and every 15 points doubles it.
What does a 20% win chance actually mean?
Across many comparable runners each given 20%, roughly one in five should win if the estimates are well calibrated. It says nothing certain about any single race, and a run of five races will not produce exactly one winner.
If the top-rated dog has the best chance, why does it lose so often?
Because the best chance in a greyhound race is usually well under half. The top-rated runner on our card wins about a third of the time; the rest of the field shares the other two thirds.