An Introduction to Performance Figures |

An Introduction to Performance Figures

The performance figures we create are the core of everything we do, but it is not necessarily so obvious to others why they are so useful. A good beginning would be to understand more of what goes into them – so here is an explanation of both What , Why , and How .

How Did It All Begin For Us - and What Exactly are Performance Figures ..?

For me, the life long passion for horse racing began with a completely random introduction to American horse racing, one late night when I stumbled upon some exciting markets, and some fascinating form information products to go with them.

This slightly backwards entrance into the Scandinavian horse racing scene left its mark, and probably still does. When I realized a few months later that there was a race course for thoroughbred horses in Norway as well, I made a sad discovery. To my great despair, the information I was now used to analyze the races with across the pond, didn’t exist at all.

To bet on outcomes without this felt like pure guesswork, and casino games has never really appealed to me. It may well be difficult to win, but impossible; no.

Fast Forward a year ahead, and I was on my way. I was going to make my own performance figures, and put them in my own form products! I mean, how difficult could that be ..?

Fast Forward a few more years, and I had the answer:

Shockingly difficult!

But I had not given up – at least not for longer periods of time – and suddenly I stood there; triumphant, proud, with my very first set of performance figures. I was a genius!

Unfortunately, it turned out so that I in my youthful hybris had done almost everything wrong, so we will need to fast forward a few more years before we had really figured it out. At this time, my little brother Gard had also been connected, and we had expanded from Norwegian races to also make performance figures for the Swedish courses, and then the Danish main course; Klampenborg. Now, the pieces really started to fall into place.

Now, our guns were loaded!

The rest of this post will be about what these numbers are, why they exist, and I will present the different schools in the field and some advantages and disadvantages with the different approaches. And, of course, a fair bit about what goes into the numbers we produce.

100 Years of Speed Figures - A History Lesson!

Horse racing and the thoroughbred racing in particular is a sport of great traditions, with hundreds of years of colorful past.

With only a few (brave!) exceptions, the concept has remained unchanged:

It’s about moving a horse faster than the competition, from one point, to another. 

A rather unbeatable concept when you think about it, but I will save those philosophical musings for another time.

At one point, perhaps 100 years ago, there were several people around who found out about the same time that in addition to assessing this competition with others when assessing the quality of performance – because you needed to do that both if you were to buy a horse or bet on it – that is, in addition to evaluating class , one should also look at who actually moved fastest  from point A to point B.

With that, the concept of Speed Figures , or Speed Ratings , or Performance Figures was born. We just call it performance figures for the sake of simplicity. The first published book to explain this very concept was E.W. Donaldson’s “Consistent Handicapping Profits”, from 1936. With performance figures one aim to work mathematically and methodically to quantify a performance – that is, to turn it into numbers – in order to more easily be able to compare them with other past performances.

Considering the problems I myself encountered when trying to figure out a sound methodology – I had, after all, 100 years of history and experience to lean on – it is probably no surprise that the first known methods had their shortcomings. It was not until 40 years later that these methods really made their breakthrough in the United States, first with Andrew Beyer’s book “Picking Winners” from 1975, which later led to the publication of Beyer’s performance figures in the Daily Racing Form from 1991 and onwards. In Beyer’s shadow side, an alternative movement based on their very own ideas about Speed Figures emerged. They went under the nickname of “The Sheets”. This movement was led by the eccentric Len Ragozin, and eventually, his former student and later “nemesis”, Jerry Brown who diverged and founded Thoro-Graph.

It was these products that first and foremost fascinated me so much, in my early flirt with the American horse racing.

We're trying to find out the true value of a horse's performance. In other words, when is a fast race really worse and when is a slow race really better?

Len Ragozin

Can't One Just Look at the Race Time, When Deciding Who is the Fastest ..?

A completely natural, yet fallible intuition

Let us first state and define that when we in this context are talking about “the fastest horse”, we are not talking about top speed, nor early pace. We are talking about who amongst a group of competitors one believe is capable of running the fastest over a full distance. That is, which horse has the opportunity to move the fastest from point A, to point B. 

The intuitive thing to think is of course to look at race times achieved in past performances. In athletics there is a great focus on personal records, for example, while in harness racing there is a lot of talk about mile times or kilometer rate. I will not say too much about whether raw times is a good measure of performance in those two domains, I simply do not know enough about that, but what I intend to say something about is that it definitely is not the case in thoroughbred racing. It is probably better to look at the clock than to not look at the clock at all, to put it that way, but there are ways to interpret these times that are just way more effective than considering raw times.

You guessed it!

Performance figures.

First of all, all race tracks are different, with differences in everything from length of the grass and “bounce” in the ground, to the size and sharpness of turns, slopes in either direction, wind conditions, entrance to turns, exit of turns, and so on.

I could continued!

A race over 1200 metres on the dirt track at Jägersro is simply not quite the same exercise as 1200 metres on the dirt track at Bro Park, nor is it the same as 1200 metres dirt racing at Meydan. 

Considering that one second in running time is about six lengths in difference – six lengths could often be the difference between a victory and a tenth place – it really does not have to be big differences before it becomes quite useless to treat them equally.

The Performance Figures Comes to the Rescue

To be able to effectively compare running times ran over the same distances, but on different tracks, you need a way to standardize these times.

The limitations of using raw times does not stop there.

Horses, for example, don’t just run over one distance during a racing career. Even a distinct sprinter could be running races over 900 metres, 1000 metres, 1170 metres, 1200 metres, 1300 metres, 1350 metres, 1370 metres, 1400 metres, and so on.

And that is on different courses, often in different countries, and over different surfaces. 

In these races, they will face horses that also comes in to the race with fresh performances from a variety of distances, tracks, and track surfaces. It simply would be a very confusing excercise to rank these horses against each other, based solely on raw times. 

We are with thoughts like these very close to discovering the performance figures, once again.

The classic performance figures, such as Beyer, aimed for quite a few years not so much higher than solving this exact problem. The idea here was that one could have “par time charts”, where one compiled and maintained comprehensive lists of expected running times for different classes on as many courses and distances as possible. Thus, one could read in the chart what a time of 1.12 over 1200 metres on one track, corresponded with in time for a horse in the same class on another track. Maybe you expected 1.11, or 1.13, for exactly the same performance on another track.

Similarly, one could do this for performances over different distances. 

To be sure, this is a clear improvement from just considering raw times, but still very far from solving all the problems with the use of running times as a measure of performance.

This book turned things upside down in American horse racing

The Problems Just Continue - But Do We See any Signs of Solutions ..?

We still have not talked about track conditions.

I do not quite know what it’s like with track conditions in harness racing or athletics, but I imagine that at least they there compete on surfaces that are more solid than what they are in gallop. The iconic Tartan surface probably has many of the same properties, no matter where it is installed, and a sulky should at least not be stuck in the ground when the starting car gets rolling.

On thoroughbred racing tracks there is enormous variety. With extreme conditions the turf track could differ with as much as 10 seconds on some distances, in plain track speed alone. This means that an elite horse can potentially do the best performance of the year on the track, but in raw time still be several seconds behind what a middle class athlete could do on a day with opposite track conditions. In this environment, with the possibility of such dramatic changes in track speed – in Scandinavia we’re even living in a part of the world where seasons have great variations – raw times becomes a very confusing concept to deal with.

Now I feel that it is way overdue to turn our noses towards those type of problems and challenges that are so difficult to address – that they provide fertile ground for many different schools in performance figure making.

In the introduction, I barely touched down on the story between Len Ragozin and Jerry Brown, where the latter parted with Ragozin to found his own company in Thoro-Graph. Their relationship remained troublesome.


A passionate disagreement about track speed!

Okay, I admit, that was probably a slightly tabloid presentation that should be elaborated in the next sentences, but it is probably also a useful illustration that horseplayers are not quite like most gamblers.

The disagreement was more precisely about whether tracks can change speed during a race day, even without obvious things such as sudden weather changes happening to them. It may still seem trivial for most people, this disagreement, but where you come down in this debate actually has a lot of implications for the methodology you’ll use when calculating performance figures.

It’s not so easy to explain this in detail without it becoming too technical for most to follow, but I will make a brave attempt without using too much of the subject terminology.

The crucial problem to be solved in modern forms of performance figure calculation, is to place a numerical value on how much faster or how much slower a course is playing, compared with it’s normal state. This should preferably be estimated on the tenth of a second. If you figure this one out, then it is easy to convert a raw time into a standardized format that can be compared regardless of the track speed. This is the whole purpose of performance figures.

Ragozin thus stubbornly claimed that courses could NOT change track speed during a race day, unless there were major weather changes or similar obvious factors in play.

His main goal of the figure making excercise was then to find one track variant (ugh, terminology!) tenth-of-a-second-adjustment for the whole day, which one could adjust all the running times that day. In other words, with Ragozin’s method, one had to come up with one just one adjustment, e.g. minus five tenths of a secocnd. Then you had to deduct five tenths of a second from all of  the running times that were registered that day. If there were races on both the dirt track and the turf track, then Ragozin would have operated with two different track variants tenth-of-a-second adjustments.

(After all, he was not so crazy that he assumed that the turf course and the dirt course had anything to do with each other in regards to track speed).

With that as a starting point, the methodology will center around methods for finding that one adjustment – and then the numbers will speak for themselves from there on and out.

The problem with placing that assumption as the main premise for the whole method, and which Jerry Brown could no longer stand, is of course that it is wrong.

Tracks change speed between races all the time!

Sometimes a little, sometimes a lot, sometimes almost nothing.

The track maintenance guys can surely elaborate in much more detail on this topic, and I would sure love to have a guest post from some of you on this blog, but I’ll try my best.

Track speed is, as far as I have been able to read up on and observe for myself, largely a question about the level of moisture in the sand and/or soil. Here there are both natural and artificial processes in play, and they are active before, in between, and after races.

A wet course from the nights rainfall will dry up at different speed depending on weather, temperature, wind conditions, etc. At the same time, the courses are often also watered multiple times during a race day. Here there may be variations in both the amount of water added, the frequency or “rhythm” of the watering regime, etc. In addition to this, various track maintenance considerations are being made both before, during and after a race day, all with the purpose of getting the course in as good shape as possible for every race. The track maintenance personnel will have an arsenal of tools and techniques available at their hands, all with possible implications for how the track plays out.

It would be very naive to assume that none of this can affect track speed. And that all of this could be observed and understood with the bare eye.

(By the way, this is a topic we would very much like to have intelligent discussions about on our forum – because there is no doubt that there is still a long way to go before I could earn my certification in track maintenance!)

tractor, sunset, landscape

Race Time Clocking, and Pace

As if this were not enough, there are a number of other challenges that will often lead to problems, if one insist on using such a mechanical methodology to the production of performance figures as Ragozin proposes. 

The first obvious challenge is about clocking, or the timing of the races. It is not only in Scandinavia that one have had difficulties measuring the correct race time, this has been an eternal problem in the USA as well. Neither is it very customary to share information out loud about when there has been problems with the timing. There could of course be taken heroic measures by the figure makers to get in control of this issue, with manual timing being the obvious solution, but then again that is not a completely waterproof method, either.

Everyone who creates performance figures wants as precise race times as possible, of course – the raw times are, independently of which school one swears to, the most important ingredient that goes into the production of performance figures. We ourselves have taken our own race times at Bro Park, for example, as the timing there has proven particularly unreliable.

In addition to ensuring that we have as precise race times as possible to begin with, we also have developed a quite unique method of calculation that makes us far more robust against timing errors, than what is common with other techniques. 

The problems with the clock is still quite trivial. Everyone would like the timing to be more precise. The really interesting source of disagreement, and where the figure makers really divide into different schools, is when one are discussing problems such as how to treat races ran at a very slow pace.

Anyone who has seen Jakob Ingebrigtsen & Co chase personal best times on the athletics field, can not help but notice how critical it is to get going at a decent pace. If you procrastinate in the beginning, you will never reach record speed, no matter how good your body feels or how much you distance yourself from the competitors at the end of the race.

I want to dwell a little bit around this example. 

The pressing question now is of course; how should we solve this issue when calculating our performance figures?

It’s the speed from A to B we measure here, don’t you remember? If they linger in the beginning it is their own fault, and to adjust for this in any way is just pure voodoo!

I have ruminated about this in an incredible amount of time, and I can fully understand that this question would be the great divider where the field divides into different schools, with different philosophies, and with completely different performance figures generated for these kind of situations. 

There are pros and cons to the different philosophies that exist. We ourselves have been searching for a compromise. 

With his method, Ragozin has no possibility to take this into account anyway, as he has married himself with the idea of changing all races with the same track variant, or “tenth-of-a-second”-adjustment. If the other races that day are run at a normal pace, and he concludes that “today we have to deduct five tenths of a second from all the race times”, then he has no choice but to do the same in the race where the first 400 metres went slow. Even though that means that the contestants never stood a chance to run times representative of that day’s true form. You will with this method have to give them a poor rating, even though the last 1100 metres of the race might have exceeded all expectations. 

But was it really a worse performance, than that third place finish from the national championships where the pace was just perfect from the start?

Ragozin would think so.

Brisnet would think so. 

Jerry Brown and Thoro-Graph do not think so.

Beyer did think so, but they later turned around, and now don’t automatically think so. 

And most importantly: We do not think so!

So, Why not?

There are two main reasons why we side with Jerry Brown in this discussion.

Class and Speed, not Either-Or!

We introduced the rationale for performance figures in the first place with a shift in emphasis from class, to speed. One went from judging the quality of competition, to how fast they ran doing it, when judging which individuals were the best gallopers. But with so much going on that affects race times, and with so many different dynamics that take place in each and every race, it became clear to us that who you beat and how much you beat them with also should count for something.

We aim with our methodology to find the perfect balance between “class”, and “speed”, and without revealing too much of our secret sauce, we do it with a regression-like exercise. In this process we use the horses’ previous performances as extra data points. Individual horses may be performing freak performances, but within a group there is probably not likely that all perform freak performances at the same time.

These data points comes in addition to the classic variables such as race time, jockey weight, and track speed.

We simply never quite bought into that premise that you would have to choose, between class or speed.

It made things much more difficult for us, of course, to not just copy an established method and accept all its assumptions and heuristics, but on the contrary work out our own methodology through a lot of trial and error. Now it just makes us even more proud about the quality we have achieved.

We also include ground loss in our performance figures, i.e we calculate the amount of lost ground in the turns into our performance figures.

A horse that runs faster than it has ever done before, but loses the race because it had to run an awful lot longer than everyone else to reach the finish line, has in any case done a great job. The form is there. The speed is there. And it is even possible to put a numerical value on the lost ground. It really just boils down to geometry.

Our post position statistics are a great place to start if you do not fully believe that wide turns hamper the chances of winning. What you see there is not about horses performing worse from the outside.

Actually, on the contrary, but I will save my theories about that for another time.

We believe in this idea about ground loss information belonging in performance figures, so much so, that we actually could be bothered to follow every single horse, through every single turn which is run throughout Scandinavia and at Meydan – and we have been doing so for almost 10 years. Give it up for Gard!!

DRF and Beyer do not.

We hereby reveal that we belong more in the “performance figure” camp of things, than we do in the “speed figure” camp of things.

I was seduced by horse race betting because it offers more mental challenge and stimulation than any subject in the formal academic world. Few people ever master it.

Andy Beyer

Fat Tony vs. Nero Tulip – Another Compromise

We now move over to the second reason why we side with Jerry, in the assessment of Jacob’s Olympic gold race compared to his third place finish in the national championships.

(I should emphasize that these are purely hypothetical examples, but for some reason I imagine that the athletics analogy is easier to follow).

One of my great role models is Nassim Nicholas Taleb, the author of e.g. “Fooled by Randomness”. Throughout the INCERTO-series, we get to know the characters Nero Tulip and Fat Tony. While Nero Tulip represents the slightly elitistic, snobbish, intellectual theorist with prospering bookshelves, an exquisite taste, and all his references in order – Fat Tony is the more street smart business owner-type of guy who through hardfought experience, trial and error – “school of life”, really – have just figured out what works, and sticks to it.

No nonsense. If it works, it works. 

According to Taleb (and many more, Daniel Kahneman for one), we humans are designed to be deceived by elegant theories, and to see what we want to see. Perhaps especially horseplayers – because what is horse betting if not the search for patterns and creation of hypotheses from at times very incomplete information? That is the intellectual challenge of this, after all! 

The point is, you have to make room for both. It’s just healthy to have a little Nero Tulip in one self, not to mention good! The ability to immerse oneself in theory and abstract concepts, or artificially elevate a good wine, or be carried away by a well-written piece of poetry – all this makes both life worth living – and is important for replenishing new impulses. That is even useful.

But it quickly becomes both unproductive and dangerous if you do not also manage to combine it with a little “Fat Tony”-mentality. 

The most important reason why we have ended up with  the methodology we have chosen – is of course because we see that it works .

The numbers are much more useful to us who uses them, when they make sense, as they do to a much greater extent if one are using horses’ previous performances as extra data points.

Or when you have methodologies that allow for changing track speeds, as they actually do change speed, even when they doesn’t seem to be. Then it’s just a matter of developing techniques and methods that could capture this effectively – without foresaking neither consistency, integrity, or reliability. It could be done. We have done it.  

There will of course not be a record breaking performance figure with a very slow first quarter of the race, for that raw times makes up too large a portion of the performance figure. But it also makes no sense to conclude that the Olympic gold medal run was the worst race the competitor had raced in a long time.

With our methodology, one is punished for fooling around in the start of the race, but not to an idiotic extent.

Our unique methodology is also the reason why our numbers are just as good on turf races, as they are on dirt races. You just cannot say that about the classic Speed Figures. Beyer are not. Ragozin are not. Brisnet are not.  

We have tried and we have failed, tested and learned, found compromises where others have parted ways, and just been really innovative in the development of a methodology that gives room for both reason, reliability, and validity.

There are strict rules for what one is allowed to do and what one could not do, but it’s still room enough left for common sense to come in handy. 

I think we have found the perfect compromise between Nero Tulip, and Fat Tony. 

I really think we make the world’s best performance figures.

Hva Fat Tony gjør som hovedbilde til denne teksten - må du nesten lese deg helt ned for å finne ut!
Meet Fat Tony – a surprising hero in this story

Now You May be Wondering What Use YOU Could Have of These Figures..?

We ourselves use the performance figures primarily to find great bets, and they usually show up in races where we think past performances were better, than what the rest of the market think they were.

A small thing, of course, but we have found great delight in participating in the popular tipping competition #DWCCTippingComp on Twitter. This is a competition that is arranged every year for the Dubai World Cup Carnival races at Meydan, lasting from january to the closing world cup day. Every year we attend the fierce competition with vigor, and we do it well. Gard won it in 2020 solely from making decisions based on the numbers we produce.

A small, little proof – perhaps – that our numbers are the real deal. (albeit, anecdotal)

We generally assert ourselves well in the Scandinavian betting markets, too, which is no small feat given the high takeout and minimal betting turnover.

The mentioned betting turnover is probably too low overall in Scandinavian racing for us to be able to actually live just off our betting, and the season is not long enough and the races are not many enough in Dubai either. But if our Hong Kong project we’re currently only tinkering with should manifest one day – it takes many years to build up a quality database using our methodology – then just maybe.

But the figures are not just tools for the bettor, they have the potential to provide great benefits in almost every link of the food chain of Horse Racing.

Here are just a few potential uses we have come up with – there is a lot of room for creativity in how to use the numbers:

  • A thorough analysis of whole racing careers and past performances, when purchasing and selling established race horses
  • Analyze horses’ strengths and weaknesses, to be used in training and career management
  • Form analysis of your horse!
  • A new weapon in your arsenal when deciding on race tactics
  • Stable form analysis, and the discovery of global stable trends
  • We could be used to recommend appropriate trainers based on horses’ strengths and weaknesses
  • A thorough analysis of whole racing careers and past performances, to be used when deciding on stallions and mares for the sake of breeding race horses
  • To find favorable races to aim for, both in Scandinavia and at Meydan
  • Identify explosive patterns, spent horses, risk of bounce and 2yo-foundation in good sheet theory-spirit
  • Identify soft spots in the current competitive field, and discover what kind of category of horses one should invest in for the best ROI
  • We could quality assure or assist in deciding the official ratings of the handicapper
  • We could assist trainers, owners and jockeys in uncovering track bias before anyone else on the day
  • We really have information that should be used in getting more fans interested in Horse Racing, and to design better gaming products
  • We could provide stylish Career Records as memorabilia from horses you or a someone you know have participated in
  • Our numbers could be used as a guiding metric when researching and experimenting with different training regimes
  • Or what about anti-doping work..?😉 We do have some examples of what cheating looks like on our numbers..

These were just some ideas, but there are more! If you have other ideas, feel free to get in touch .

Number of horses in the database
Number of races assessed
Number of performance figures given
Number of turns analyzed

Check Out Our Top List from 2013 to Today!

Doesn't this make good sense, maybe..?

Play Video

Want to know even more? Nothing pleases a passionated rating heart more!

Here is perhaps the internet’s best explanation of the different ways to calculate performance figures:
In addition, we encourage you to ask questions and start discussions on our forum – we even have an English speaking-subforum!

Leave a Reply

Your feedback is valuable for us. Your email will not be published.

Please Wait...