Is it possible to know what a horses true chance of winning is?
Whatever anybody tells you, the answer is no.
We can estimate a horse’s true chance of winning, but we won’t ever know it.
When I first started betting on horses, I was told to stop trying to get the exact probability of a horse’s chance of winning because you would only ever know it after the race.
And… there’s only one true chance.
The winning horse has 100% chance of winning and all other horses had a 0% chance of winning.
It took me a long-time to get my head around that, but now that I do, if you’ve got any questions on it, please leave me a comment and I’ll answer them.
However, it’s common knowledge that the odds just before the race begins are a very accurate estimate of a horse’s chance of winning,
This is something which is very easy to test.
Take every horse with odds of 2/1, and you should see around 33% of them win.
Odds of 2/1 indicate that the horse has a 33% chance of winning.
You can calculate what a horse’s odds estimate it’s chance of winning by dividing the number one by the decimal odds.
For example…
A 2/1 horse is decimal odds of 3.00. That means:
1 ÷ 3 = 0.33 or 33% chance of winning
But to save you from doing it, I can assure you that over the long-term the pre-race odds predict the chance of a horse winning accurately.
Creating an odds line, like we do in RA Pro with our PR Odds, is a process of trying to predict a type of horse that will win a certain amount of time over the long-term.
For example, we’d expect all of our 5/1 shots to win 16% of the time in the long-term.
Why do we do this?
Because it helps us to know when to place a bet and when not to place a bet.
This is how…
If we know that over the course of a year our 5/1 selections win 16% of the time, then if the odds with the bookmakers for one of these selections is higher, we know that we’ve got a value bet.
Bet these selections every time they come up and you’ll make a long-term profit.
In Summary
It’s not possible to ever know a horse’s true chance of winning.
But we don’t need to know a horse’s true chance of winning in order to make a profit.
All we need, is to know that if our odds line says a horse has a 16% chance of winning, that those horses will win 16% in the long-term. That doesn’t mean one specific horse has that chance of winning, as I’ve mentioned, there’s no way of knowing that. It means that all these horses over the course of a year or two will win 16% of the time.
With this information, we can easily make a decision on whether to bet on a selection or not, simply by only placing the bet if the odds with the bookmakers are higher than the odds in your odds line.
I disagree that the correct probability line before a race is 100% on the winner and 0% elsewhere, simply because the information about the race unfolding doesn’t yet exist.
You wouldn’t say that the probability of flipping a head is 100% or 0% before it was actually flipped – it’s 50% until it is flipped, and as it goes through the air, the probabilities change one way and another until it finally settles.
Same with a race – a poor start or a stumble affects the win probability of each of the horses as the race unfolds.
The 100% line is only correct _after_ the race.
Colin you are right, the 100% line is correct at the end of the race and I agree that as the race is being run the probabilities change depending on what is happening during the race. What I have found is that a lot of people think that there is only one correct probability for a horse at the beginning of the race whereas there are lots of correct probabilities depending on how you are creating the line. If you can accurate line that uses different information to the public one then you are in a position to start obtaining value.
Ah! How do we assess a horse’s true chance of winning a race?
This is the big question to which there is no easy answer.
There are so many factors to consider before the race ( form, course, going etc.) and so many consequences in running ( interference, poor start, jumping errors etc.) and other factors that we punters cannot possibly know about (sudden improvement, trainer’s plans etc.)
But all is not lost because if it was a simple formula everyone would know and that would be that!
Because we can all formulate a different solution we can get an edge and obtain value when conditions are right for our individual assessment.
This requires hard work using ratings and other factors to create a comparison between horses that is represented by a numerical value or forecast price. If the actual price on offer is greater then over a series of bets a profit will be made.
It is all about probabilities, averages and long term results and each person who manages to produce a reliable system of analysis may well back different horses in the same races but still make a profit in the long term.
Sometimes it is best to keep things simple since the game is fraught with unexpected results which, after many hours of hard work, have you tearing your hair out!
I developed a simple approach for non-handicap races many years age using the good old Daily Mirror.
This involved giving points for three factors.
1) First four horses in the betting forecast (4,3,2,1)
2) SF, F or EW (3,2,1)
3) First three Spotform ratings (3,2,1)
Each horse will have a rating. Divide this by the total points to get the probability.
Example: If the horse was Forecast favourite, SF and Topspot the total score is 10 (4+3+3).
If the total points for the race is 20 the probability is 1/2 or 50%, an even money chance.
Should the price on offer be 2/1 then value is obtained since we will back on average, one winner from two.
A very simplistic approach but it illustrates the principle without delving too deeply into form and uses factors which have proven probabilities.
Thank you for this excellent post Roger, and for sharing your approach.