Wouldn’t it be great if all we needed to be profitable was a single rating, we always bet the top selected and counted our money.
Over a decade ago, that was what I thought would happen. Which is why I completely understand when people ask me which of our ratings will make them a flat stake profit by betting the top rated.
I’m proud to say that quite a few of our ratings will be break-even at BSP under specific race conditions. However, if I had to pick a single rating from our software, and use nothing else, it would be the PR Odds, which is this one here…
This is a power rating, or combination factor, which means it’s a combination of other factors. The name itself isn’t important.
What I want to share with you today is the basics of how we create this factor so that you can make your own.
There’s a common myth that you just need to find a few ratings which work well, and then you can use them in every race.
Nope, not true.
Well… in theory it is true, but they have to be incredible ratings, probably all power ratings, and there has to be very few people using them.
The approach I use, and suggest you do to, is different. It works like this:
- Choose a set of specific race conditions
- Find the ratings which are most effective in these conditions
- Combine any similar ratings to make a power rating
- Use these to find your bets
Just four simple steps. Nothing complex, but incredibly powerful.
Let’s start start with the first step, this is the one that most people skip. It doesn’t matter how many times I write about it, suggest people do it, the majority of bettors skip over it. They think it’s not necessary.
But the truth is… it’s probably the most important step out of all four!
- Choose a set of specific race conditions
In the UK and IRE there are a myriad of race conditions. We’ve got different distances, ground conditions, courses, bends, horses.
And when I say horses, I mean the variety here is huge. If you go somewhere like Hong Kong the number of horses is surprisingly limited, which is one of the reasons the Benter and similar betting teams started there. But here, the pool of horses is huge.
It’s this variety of conditions which makes the racing here the best in the world.
However, when you combine this with the general lack of information about horses racing, it makes it amongst the hardest country in the world to find profits from.
There’s a reason that the big whale teams have never come in here!
If you’re thinking… hang on a second Michael, there’s masses of information, you have hundreds of ratings for a start… then you’d have a point. Let me clarify!
When you compare the amount of data we have on horses in the UK with other countries such as the USA, Australia and Hong Kong, it’s pitiful. In most countries the trainers have to report how the training runs went, any operations and medication have to be openly declared, every single element about a horse and it’s connections is publicly available, and the tracks have sectional timings so you can get accurate pace information.
We just don’t have that kind of information here.
In order to try and balance the equation back to our favour, I have spent nearly a decade developing ratings to measure almost every aspect of a horses performance that can be measured, and to do that in multiple ways, to try and fill in as many of the blanks as we can.
This is why choosing the race conditions you want to focus on is so important.
Once you’ve got the race conditions, for example All Weather handicap sprints up to six furlongs on Standard ground (you need to be this specific, even more so!), then you can start on step two.
2. Find the ratings which are most effective in these conditions
Finding the ratings that are going to be most effective. This part of the job is going to take up the majority of the time it will take you to create your own odds line. But it’s going to be worth it.
There are a number of ways that you can determine the effectiveness of a rating, I’ve tried most of them, and the one I usually go with is amongst the simplest.
Pool Impact Value, or AE as it’s often known.
This is simply a measurement of whether the rating finds winners more often than the odds suggest it should. This blog post explains how to calculate it.
Here’s how I do it:
Find all the races that have similar conditions to the race you want to analyse.
Once you’ve got all these races take your first rating, and split it into ‘buckets’. These buckets are simply a way of grouping a range of rating scores together. If you’re using rank ratings, which is the simplest, then you’ve got 1, 2, 3, 4 etc… with 1 being the top rated, 2 being second top rated and so on.
If you’re using a raw rating, where the score could be anything from 0 to infinity, with decimal places and everything in-between, it’s a bit harder. The best way to break these into buckets is to take all the horses in your sample for that rating, and try to split the buckets as evenly as possible, so there’s a similar amount of horses in each bucket throughout the sample.
When you’ve got these buckets you can move on to step two…
Calculate the pool impact value for each bucket.
Following the instructions in the blog post showing you how to calculate the pool impact values (PIV’s), create them for each of your buckets.
Rinse and repeat this for each of your ratings.
You now have a mass of data. Congratulations. Once you have a mass of data you can start to use it to determine how to find winners!
Kick out any ratings that don’t show much predictive strength.
Again, there are a huge number of ways to do this, for simplicity I would recommend simply removing any where the PIV’s for a rating are all gathered around the score 1.00.
You can only judge this on a case-by-case basis, but if out of your ratings you had some which had PIV’s going from 0.30 to 3.50 and some which were going from 0.95 to 1.05, then I’d get rid of the ones around 0.95 to 1.05.
On the other hand, if most of your ratings were from 0.95 to 1.05 and some were all around 0.99 to 1.01, then I’d get rid of the 0.99 to 1.01 ratings.
You are looking for the ratings which look to be the biggest predictors. It doesn’t matter whether the horses win far less than expected or far more than expected, both pieces of information are useful.
When you’ve completed this you can move on to step three…
3. Combine any similar ratings to make a power rating
If you’ve got ratings which are very similar in what they measure, you can often create a better rating by combining them together to create a single rating. Here’s a blog post showing how you can do that.
This isn’t a completely necessary step, but it can improve your odds lines. It may be a step you want to skip initially, and then come back to at a later date.
Which brings us to step four…
4. Use these to find your bets
You’ve now got the ratings, and the PIV’s for each of their buckets, which you can use to make an odds line and find your bets.
I should point out that you don’t have to make an odds line, that’s just one way of doing it. You could use the scores for each horse without turning it into an odds line.
Heck, you don’t even have to make scores for each horse. You could simply score each horse like this:
- Look at a horses rating
- See which bucket this rating falls in
- Note down the PIV of the bucket for that rating and put it next to the horse
Do this for every rating and every horse, and you can simply use your own ability to look at these scores to make a judgement on which is going to be the best horse in the race.
If you find that too much, you can combine these PIV values together by multiplying them. Multiplying them together will create a single factor which you can then use as your master factor to determine which horse is most likely to win.
To turn this into an odds line you could use this approach. Like everything, this isn’t the only way to change your factor into an odds line, but it will do the job nicely.
Once you’ve got your odds line you can look for value bets, or use a combination of the contenders from your raw factor and the value odds found.
To make this approach most effective you should choose ratings that are as independent from each other as possible, and have as few as possible. Just five or six ratings will be fine.
In other blog post I will look into how you can determine whether your ratings are independent of each other, or whether they are measuring the same thing, even if you think they’re not!
If you’ve got any questions on the above process, which would be only natural, leave me a comment below.
And if you’d like to have this scoring done for you, on every single race, dynamically calculated, with a splash of artificial intelligence thrown in to optimise it, all at the click of a button…
Then you’ll want to join the Race Advisor Pro Members Club here and start using the PR Odds factor on every race card.