What are subgrades and how do they work?

An analysis of the impact of subgrades for horses. Grades don't tell the whole picture, we need to dig into the subgrades to better understand horse performance in PFL.

Petrocker

6/1/20248 min read

It's easy to get wrapped up in horse grades. They are the most obvious thing about a horse when you look at the PFL website. But as @NFTLeano has explored in his YouTube video, grades obscure a variety of detail that is going on beneath the surface. Let's try to add something further to the exploration of this topic and help stables better understand their horses and what to look for when buying or claiming, breeding or selling horses.

How much difference is there between grades?

Anyone who has played PFL will tell you that it isn't surprising when an S grade horse beats a field of S+ horses, or even A+ can, on their day, beat S horses. It's a bitter pill to swallow when you are the owner of the S grade horse, but it happens all the time. Why is this the case?

Well let's take a look at benchmarks, which is a normalised measure of recent horse performance. As players we don't have the specific rules of how the benchmark works, but broadly better recent performance drives up your benchmark, and poor performance drives it down. Its main use is for benchmark races that limit who can enter based on current benchmark, regardless of what grade your horse is.

The above chart plots the benchmark for every horse in every race and what we get is the distribution for each grade. First thing to notice is, apart from the B- and B lines, where horse sample size is small, each line is smooth. This suggests to the naked eye that benchmarks are a normalised function - that is performance is mapped to a normal distribution (bell-curve). The second thing to notice is that there is a high degree of overlap between grades. S grades are not universally better than S- grades for example. The lines intersect at roughly what I would call the 60/40 area, where the bottom 40% of one rank are below the 40% of the lower rank.

Grade A+ horses have a median of 63, S- 66, S 70 and S+ 74. If your horse is one of those grades and commonly has a benchmark above that level, its probably an above average horse for its grade, and obviously the further to right it goes, the better it is. We have looked at similar analysis using normalised finishing times in this blog post with some similar conclusions that there is plenty of overlap between grades.

Obviously horses' benchmarks will vary after each race, so it's about finding the range within which your horse operates rather than a single figure. I think a chart like this helps us players quickly assess whether a horse we see in a claimer or on the marketplace is good for its grade. It's a shortcut or a ready reckoner, not the definitive answer, but useful nonetheless. It's one piece of the puzzle, but canny players should combine these insights with Gap Data's https://gapdata.racing/gap-ranks/ or PhotoFinishEdge's FF2 rating to further dig into a horse's ability and conduct their own analysis.

So why do horse benchmarks. or normalised finish times overlap? Why do we pay such a premium for a S+ colt over an S colt , when there is such an overlap? To understand this we are going to have to do some sums (groan), and I'll try to keep this as simple as possible.

Understanding subgrades within grades

Each horse has a grade that is clear for all to see and six attributes that are hidden until retirement, in addition to the three preferences we can see for course direction, surface and condition. This makes up the profile of each horse. This analysis is performed purely on retired horses, because we need to see their attributes. The six attributes are start, speed, stamina, finish, heart and temper. Each of these six attributes are graded in a similar fashion to the overall grade.

Shockwave is an S grade horse, with 10MM derby earnings mostly from breeding, and you can see why. An S grade horse with 9 stars and several attributes above his baseline grade of S. He has an S+ for speed and an SS- for stamina along with an S+ for temper, all above an S. In fact he looks closer to an S+ than an S. In order to do some basic math(s) we are going to have to turn these letters into numbers. We do it using the following table:

So going forward an S- is a 13, and an S is a 14. For Shockwave it was 3 S grades, 2 S+ and an SS-. On our number line, we know that an S+ is worth +1 more than an S, and an SS- is worth +2 more than an S. So Shockwave is not just an S, but a +4 S, because he has 4 points more of grades than an average S.

An average S has 6 x 14 = 84 points. An average S+ has 6 x 15 points = 90 points. Shockwave has 88 points, so 2 points below an average S+. Let's look at another horse, Morwen.

Morwen is an S+, with 3 S+ grades for start, speed and stamina, an S for finish and heart and an S- for temper. If we convert these grades to numbers we see that Morwen is a -4 S+, with 4 attribute points below an average S+ because finish is -1, heart is -1 and temper is -2. If you add up all the points Morwen has 86 points, two less than Shockwave who is an S grade horse. Purely from an attribute point perspective Shockwave has more points than Morwen despite being a lower grade horse.

How can this be? Well there are two theories here. The first is that there is another level of detail we can't see in the letters - the equivalent of there being one decimal point that we don't see. An S is not just equal to 14, but might represent >= 13.5 and <14.5. That way Morwen might have strong S grades for finish and heart (say 14.4), whereas Shockwave might have weak S grades for finish and heart (say 13.5), so in reality there is no overlap. We can't see this hidden layer. If you play Football Manager, player attributes can be displayed on a 1-20 scale but in reality the game uses a 1-255 scale behind the scenes. The same could happen here. The second theory is that grades actually overlap and a great S is better than a poor S+. I think this requires some more investigation. If you have theories and want to share them there is some discussion on the theories channel of the PhotoFinishEdge Discord server.

What I would say is that before you buy a retired horse on the marketplace, quickly figure out whether its a plus or minus horse. Does it have more points for its grade or less? A horse like Shockwave is closer to an S+ than an S, Morwen is closer to an S than an S+ (using our number conversion). Some analysis of distribution of these +/- figures reveals that +/-5 is the limit. Few horses have +/-5 points from their grade. Clearly six points +/- forces a horse into a higher grade. In the data used in this analysis only 7 horses have achieved +/- subgrades, for example Filly Cheesesteak, is an -5 A.

This is a table of all the retired horses in the analysis, split by their over/under score, compared to on track performance. Firstly, the numbers seem fairly normally distributed - so this could well be a random process, rather than something that could be contrived through breeding strategies - however we will need to do some work to conclusively prove this one way or the other, which would be another article entirely.

Secondly, just looking at the colour coding we can see that +4 horses win twice as often as -4 horses. It's not black and white, we are not factoring in preference stars and we are mingling grades together - so there is a lot of noise in there - but still the pattern is fairly obvious; horses with + subgrades are better performers (all other things being equal) than horses with negative subgrades.

My advice would be if you are selling or offering stud fees make sure you include a shoutout for your +subgrade horses, and keep quiet if they are negative. You should definitely charge a premium for higher subgrade horses when you sell. A +4 horse is equivalent to a -2 horse from a higher grade (same number of points using our number scale).

The below table looks at just S grade horses, removing some of the noise to show the same pattern exists within a specific grade.

The next table looks at the role of preference stars on win percentage, and as expected stars compound the pattern. Having more stars and more + subgrades will lead to more on track success. A 9 star -4 subgrade horse wins at 12.6% and a 9 star +4 wins at 22.2%. There is a clear difference. The rest of the table just illustrates how hard it is to win without high preference stars.

Down the rabbit hole of subgrades

So today we have explored that there might be an overlap between grades of horses, which explains why lower grade horses commonly beat higher grade ones. We now know the boundary of grades is +/- 5 subgrades, so a horse can have a maximum of 5 subgrades above or below its overall average. Those with 3 or 4 positive subgrades win more often (go figure). A +3 horse is half a grade above the average - this might be important when you think about breeding options. You might want to avoid breeding with -3 horses who are half a grade below their advertised level. If you remember nothing else, remember the subgrades.

The success of a horse on the track can be somewhat explained by preferences and subgrades but it's the configuration of these grades that actually predicts success at specific distances. For example, to be successful at short distances horses need start and speed. We will look at this in a future post, which should add another dimension to this analysis today and build on our knowledge.

There is an argument to say all of this is moot; we don't get to know our horses' attributes until they retire. This is clearly true, but understanding how the game works, gives us a small edge, which when combined with other small edges should hopefully make us all better players.

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 your spend your own money. Nothing I write about should be considered financial or investment advice.

What are subgrades and how do they work?

How much difference is there between grades?

Understanding subgrades within grades

Down the rabbit hole of subgrades

Contact

Social