Generating more wins

Everyone wants to win more often, but sometimes we can't see the wood for the trees. A short post on the importance of preference stars and subgrades on win rates.

ai_paddock

3/1/20257 min read

Yes PFL is complex and deep yadda yadda yadda, but imagine for a minute we made it really simple. We strip away all the folklore, the PvP misinformation, vested interests, bagholders, false prophets (I put myself in this camp FYI) and naysayers. What do we want? To win races. How do we win more races? Do the obvious stuff.

How can I win more races?

There are two obvious ways to win more races. Breed, buy or claim horses with high preference stars. Breed, buy or claim horses that have positive subgrades. The first one is easy to say and easy to do. The second one is harder to define and harder to achieve.

Horses with low stars don't win very often. "But why don't 9 star horses win 25% of the time if stars are so important?". Stars are a relative measure. Having 9 stars isn't much of an advantage if all the other horses have 8 stars. Equally, if you are in an archetype where the number of stars is fairly low at the top grades, then having 6.5 stars in say LTF is like having 15 stars in RDF. A chart like this can't tell you the relative importance of stars. What is an extra 0.5 stars worth to me? That's the question you should be asking yourself when you throw all your breeding money at 4k SS- studs with 4 stars. Maybe when we get SS- maidens, 6 stars will be a champion. Only 4% of S+ horses have 8 stars or more. As people rush to breed with low star minimum cost SS- studs, stars are being diluted, star quality is getting worse. This is an opportunity for the canny stable owner.

This pattern in the above crosstab holds true across every grade and archetype, the numbers change but the pattern doesn't. But this shouldn't be a surprise to anyone, in fact it should be fairly obvious, the game literally tells us that this is true. You can't have enough stars.

Star performers

If we take a cohort of S+ horses, who have run in the last 6 seasons, we can analyse their performance on the track purely by the number of stars they have.

If you ignore the tiny sample of 9 star horses here, there is a clear bad to good, red to green, pattern across a number of useful measures. Not only do high star horses win more often, they beat more of their opponents, they get (slightly) higher morning line (ML) prices, they make a profit in Derby, and probably most importantly they win more restricted stakes and open stakes races per foal (final 2 columns). They are 10x more likely to win the big races.

Unless you are playing 5D chess, intentionally breeding horses with fewer stars is a bad move for us mere mortals. No archetype has more than 5% of S+ horses with 8+ stars. If you can't access the best bloodlines or lucky (some call luck the winning gene) shilled horse, you can always work on increasing the number of stars.

At the time of writing, towards the end of week 3 of season 25, the collective, now juvenile, foals of one 3.5 star SS- stud currently have 48 runs and 0 wins in season 25 so far.

Subgrade success

Subgrades are kind of the opposite of stars. We can't see them until retirement, it's more difficult to breed for them, and even if we have a +4 subgrade horse it might have a bad combination of incompatible attributes like start and finish. However, that doesn't mean that they aren't equally as important as stars.

Taking the same cohort of S+ horses born over the last 6 seasons, and looking at those that have retired we can analyse the same data by subgrade. Interestingly, at least to me, is the lack of positive subgrade horses in the S+ grade, even in season 25 (only horses born season 22 and before can be retired atm). While the S grade horses have an even mix of negative and positive subgrades, only 10-15% of S+ horses have a positive subgrade. So there is some way to go to fill out the S+ grade. This also makes it more likely to be able to win with a negative subgrade horse. Just like with stars, attributes are relative to your competition and if the competition is 80% negative subgrade horses, you don't need to be a massive subgrade horse to win. Clearly the SS- grade is even more skewed to negative subgrades at the moment. This is also why good S horses can sometimes beat out fields of S+ horses.

As with the previous table, red=bad, green=good. Horses with higher subgrades win more often, are profitable and win a higher share of both restricted grade and open grade races, although given the skew in population distribution if you go look up a major winner, it might well be a negative subgrade, maybe got lucky on the day, just because +subgrade S+ grade horses are few and far between.

As previously stated, it's not enough to have positive subgrades (which you won't know until retirement), you also need your strengths to be in complementary attributes, like start and speed, or stamina and finish. Having gaps in your profile is bad (e.g. start=S+ speed=S- stamina=S+ finish=S-) - it can mean your horse is bad at all distances.

If you take a look at the "ml_beaten" column you see much stronger differentiation than the stars table. ml_beaten is calculated by looking at each horse and determining what % of horses faced did the horse have a better ML price. If your horse was the favourite it would have beaten 100% of rivals in that race. Take that figure and apply it to all races - that is ml_beaten. You can see strong positive subgrade horses, like say +3 have an ml_beaten% of 81.2% whereas a -3 has a figure of 37.8%. From the earlier table a 7 star horse has 52% and a 2 star horse has 49% - not much difference.

This lends weight to the idea that attributes are baked into morning line prices more than stars. But, don't take my word for it, take a look at the scatterplot below. Right to left is what % of horses a horse has faced and beaten (kind of like average finish position - but better because it takes into account field size), where the further to the right the better the horses' results are. On the vertical axis is the aforementioned ml_beaten, with horses at the top more fancied by ML prices and towards the bottom are horses that are more likely to be outsiders. Overall you can see there is a somewhat weak relationship with a dominant diagonal from bottom left to top right which suggests that some element of actual performance is captured but not accurately (r^2 is about 0.2). The dots represent individual horses and the colour of the dot represents the number of stars.

If you squint you can see the dots on the left are more red and the dots on the right are more dark green. This suggests that the ML algorithm over estimates horses with low stars and under estimates horses with high stars. If the model was 100% accurate all horses would be on a diagonal line from bottom left to top right. If you have a horse with high stars and is ML favourite - that is a very good sign, If you have a horse that is an ML outsider and it has low stars, maybe rethink your breeding strategy.

I recently discussed this very topic on Back in Blood's Ponies on Demand podcast which you can enjoy me umming through, discussing ML prices in more depth.

So, the moral of today's post is be mindful about preserving stars when breeding. The shortcuts to the top often pass through low-star territory, and the consequences of not getting the SS- breed that you wanted is potentially an unraceable and unbreedable S+. Horses with higher attributes win more often. Your job, if you choose to accept it, is to guide your foal toward having +subgrades with a combination of start/speed, speed/stamina or stamina/finish, and at least average character (temper and heart). This is harder than it sounds, and you could just leave it all up to lady luck. If you have a massive stable you will bumble into many good horses every season. If you have a small stable and budget you need to weigh up every decision. Fingers crossed for next Foals Day!

Join the fun and put these insights into practice at PhotoFinish.Live and if you are considering starting your own stable please consider using my referral code: PADDOCK or just click on this link: https://signup.photofinish.live/?referralCode=PADDOCK

Please remember this is a web3 game where you spend your own money. Nothing I write about should be considered financial or investment advice.

Generating more wins

How can I win more races?

Star performers

Subgrade success

Contact

Social

Write your text here...