Moneyball and the 2 Year Old in Training Sales

The concepts found in ‘Moneyball’ can transform the 2 year old in training sales and greatly increase ROI for auction buyers. I have written and rewritten this post a half a dozen times – ranging from a 5,000 word opus down to just the main bullet points, so this piece may seem to be a bit scatterbrained, and I apologize in advance.

The central premise of Moneyball states that the collected experience of baseball insiders is, by nature, subjective, and therefore flawed in some respects. Rigorous analysis resulted in the creation of new statistics that better defined a valuable player via objective means, and this often flew in the face of conventional ‘wisdom’. The Oakland A’s under Billy Beane (played by Brad Pitt in the movie version) ignored common statistical measures of success such as batting average and stolen bases, and found that numbers such as on base percentage and slugging percentage were more closely tied to wins. This concept led to the identification of undervalued talent, and the Oakland franchise repeatedly won as many games as the top spenders, such as the New York Yankees, despite a payroll at the bottom of the league. (Oakland spent approx. $40 million while the Yankees spent over $130 million in any given season)

It’s fairly easy to parallel this concept to the thoroughbred business, we even have a single word defining purely subjective means of evaluating talent: horsemanship.

For decades, auction buyers have used bloodstock agents and trainers to purchase racing stock at both yearling and 2yo sales around the country. Until recently, there has been very little science/technology involved in these decisions – so how has pure horsemanship fared in selecting the game’s future stars?

(Data table taken from ‘Thoroughbred Auctions: Analyzing ROI in Yearling and Two Years Olds in Training Sales’, a study authored by J.J. Risego, Preston Guilmet, and Matt Carter – three students of the University of Arizona Racetrack Industry Program. ROI, or return on investment, was calculated using racetrack earnings from the careers of auction purchases at several major sales vs. the purchase price and training costs, which were estimated at $25k per year. No residual stud values were factored into the equation, so a few ‘home run horses’ could have improved the ROI factors by a fractional amount.)

Not so good.

If you believe that horses are mythical creatures that don’t obey the laws of exercise science and that a negative 70% ROI is the best possible result from selecting potential racing stock at such an early age – save your time and stop reading this blog. You are not my target audience and you will only get angry at what follows because it is threatening to your core beliefs. In Moneyball terms, you are ex-player turned broadcaster Joe Morgan, the most vocal opponent of the Moneyball concepts – even though he has admittedly never read the book. Is it any wonder that when he was the manager of the Cincinnati Reds that the website was formed? What a blockhead – publicly lambasting a book which he never took the time to read, but sadly Mr. Morgan’s behavior is indicative of the baseball establishment’s feelings.

So, what is overvalued when it comes to future thoroughbred racing success? Pedigree and subjective opinions of quality based on horsemanship are the main culprits.

At the yearling sales most of what is used to influence the buying decision is pedigree, the walk, and the overall subjective appearance of the young colt or filly in question. At least, that was the case before 2005 when this data was compiled. Since then, more objective scientific means have come upon the scene: heart size and gait/biomechanical analysis. But, each of them has a significant hole as well.


Here is the formula that heart size practicioners rely on: cardiac output equals heart rate times stroke volume, or CO = HR X SV.

Stroke volume is further defined as the size of the left ventricle times the amount of blood pumped with each beat, or ejection fraction: SV = LV X EF. Consequently, the amount of blood available to provide aerobic energy (CO) is the product of HR, LV, and EF. This is neither a Thoroedge formula, nor a horse racing one: this is page one of any exercise physiology textbook and it applies to horses, humans, camels, greyhounds, rats, etc. It is science, it is fact, and it is not open to interpretation.

LV and EF are quantified at rest, but no one ever measures HR during exercise! That is but one reason why you can have a huge heart in a terrible runner. Experience tells me that as many as 15% of the offspring of our top US stallions and accomplished racemares have maximum HR values of just 205bpm, well below the average of 230bpm. No horse can overcome this 10% deficiency in CO. Here is the key: you can measure LV and EF at rest, therefore you can theoretically cover all 4000 yearlings at the KEE sales, but you cannot get maximal HR in a yearling at a walk or jog. However you can get maximal HR in a 2yo who is actively training for a breeze up sale. More on this shortly….


For our terms, we’ll refer to biomechanics as simply conformation: a set of measurements/angles of a horse at rest. Problem number one is the ‘at rest’ state. No numbers taken at rest are going to be hugely predictive of athleticism at 35+mph. Some will, but many won’t. Gait analysis solves this problem at the 2yo sales by analyzing many of the same concepts, but during race speeds. Excellent improvement, but still we are looking at the outside of the horse and quantifying efficiency of movement – yet we leave out the intensity of effort, which is the other 50% of the equation. We need more useful data and in 2011 it is available to us, albeit with some extra effort.

The overall concept behind Thoroedge: anything that happens of importance in a horse, happens during gallops and breezes. Moneyball was quick to point out that selecting a baseball prospect in high school was much more likely to result in failure than selecting a college ballplayer. Likewise, it would seem that the thoroughbred 2yo sales would give us better ROI numbers than the yearling auctions, however the chart above proved that to not be the case. Heart size and gait analysis are very much on the right track – but the missing pieces of info from these methods can be available to us at the 2yo in training sales.


Like the Oakland As, we must create our own statistics to maximize ROI and here they are: Maximum Heart Rate, V200 and Heart Rate Recovery.
All of these numbers are captured during the conditioning process for a 2yo headed to the in-training sales via onboard HR/GPS equipment, but not necessarily on the day of the public breeze itself.

Maximum HR is genetic and is unaffected by training. The average for thoroughbreds is 230bpm, and I have seen many successful runners worldwide with max HR values as low as 215bpm – but none with a number of 205bpm, those rarely even make it to the track in one piece as every step they take on the track requires 10% more effort. This number is what I call a ‘low hurdle’; just passing this test with a number of 225bpm or so doesn’t guarantee success by any means, but failing this test does guarantee failure. As a matter of fact, any horse with a max HR of 205bpm or lower isn’t even subjected to the following two measures of athleticism, as he belongs in an equestrian event rather than a thoroughbred race.

Once we know maximum HR, we can calculate V200: velocity of gallop at a HR of 200bpm, which on average is 85% of maximum. Todd Pletcher and a few others unknowingly call this ‘cruising speed’, and they are right on target. Heart size alone is meaningless for our purposes, what I want to know is how much work that heart can fuel – and I don’t need 20 years of data and a massive sample size to understand that ‘faster is better’ when intensity of effort is controlled for. Poorly conformed horses will often have a poor V200 unless some other factor makes up for the deficiency. Even very well put together horses will suffer with respect to V200 if any other factors are lacking such as: capillary density, mitochondrial density, poor neuromuscular control, oxygen levels in blood, respiratory difficulties, etc. I am not a vet, and I don’t know what is right or wrong with a horse, but quantifying how the whole system performs is well within my wheelhouse.

Here is V200 data I have compiled over the years in horses actively campaigning in the US over dirt:

V200                      Class

<20mph               can’t win, likely unsound or simply unathletic
22-24mph            not likely to hit the board, but in the ballpark
25-27mph            $25k claimer at lesser circuits, maiden winner at KEE or CD
27-30mph            allowance level up to state bred stake
30-32mph            getting closer, possible G3 black type if placed correctly
>32mph                graded stakes athlete, 35mph is your superstar

Horses derive energy in a race from mainly two systems: aerobic and anaerobic. V200 measures aerobic efficiency by collecting HR and speed data during a routine morning gallop around 2:15-3:15 min/mile pace. When they start to breeze and use anaerobic energy in the absence of oxygen, well then we need this last measure of ability/performance…

Heart Rate Recovery data is collected once the horse hits the wire after a breeze and begins the gallop out process. During this time the heart rate will drop precipitously from the maximum HR achieved during a piece of fast work like a 10-12sec furlong. Ideally heart rate will be 120bpm after 2min of passing the finish post, and 80bpm another 3min later. That would be 100% recovery and is a home run for a 2yo in the midst of an aggressive conditioning regimen.

Take two runners at a breeze-up sale. One works an eighth in 10 flat, while another comes across the line in 11 seconds. The faster of the two may show a recovery HR of just 62%, while the ‘slower’ one exhibits a 91% recovery. This is significant. As the old saying goes: ‘It’s not how fast they go, but how they go fast’ – and this is what that maxim looks like in actual numbers. Speed is overvalued in these instances, whereas stamina is a relative bargain. Neither of these horses is ever going to run a 10sec furlong again, and even the fastest horses are only at peak speed for a few strides in any race. Stamina wins races (even at 6F) and is often overlooked at auction, giving us a prime buying opportunity IF we can collect the data from inside the animal in question.

**To emphasize: we are not looking at the potential of stamina in a pedigree, we are gauging the actual presence of stamina during track exercise – big, big difference.


Beyer speed figures, the Ragozin sheets, Thorograph, etc. are all statistics put out to help handicappers pick winners. Sometimes they work well, sometimes they don’t, but they all share the same flaw: they only quantify workload, not intensity of effort from each individual horse. Too many in this industry have been overwhelmed by numbers purporting to define class or ability that don’t come from inside the actual horses in question. This reminds me of another old horseman’s saying: ‘People have opinions, horses have the facts’ – so very true. D. Wayne Lukas I believe is the originator of the phrase?

Both of these old sayings prove that horsemen and women have been considering the scientific side of the sport for decades, albeit unwittingly. I intend to bring these concepts to the forefront. Science and technology has vastly improved many facets of the industry over the past few decades – pretty much everything but equine selection and conditioning – and that is an oversight we can use to gain an advantage in today’s marketplace.

Look, pedigree is merely potential on paper – once the foal hits the ground, we can now measure whether or not the preferred genetic traits are being expressed in terms of actual performance. Pedigree doesn’t matter again until one hits the breeding shed after retirement. Both baseball and horse racing are rooted in tradition, but there has now been proven to be new knowledge available to baseballers, and there is also new horseracing knowledge as well.

In conclusion, I leave you with this quote directly from Moneyball – where you can substitute ‘horseracing’ for each reference to ‘baseball’:

“When you think of intellectuals influencing the course of human affairs you think of physics, or political theory, or economics. You don’t think of baseball, because you don’t think of baseball as having an intellectual underpinning. But it does; it had just never been seriously observed and closely questioned, in a writing style sufficiently compelling to catch the attention of the people who actually played baseball. Once it had been, it was only a matter of time—a long time—before some man of action seized on newly revealed truths to gain a competitive advantage.”



About bpressey

Equine Exercise Physiologist

Posted on November 11, 2011, in Uncategorized. Bookmark the permalink. 15 Comments.

  1. Well done Bill. I have made some notes on related topics, with a few of the scientific references that are relevant to this topic.



    David Evans

    • Hi David, nice to hear from you. Neat post coming up from a young vet student turned trainer in Germany who used info on this blog to hit the money in his first 8 starts: 2 wins, 3 places, 2 shows and $25k in purse money. Stay tuned….

      With regards to the 2yo in training data, many here are doing the heart scans and gait analysis – together we hope to add the HR/GPS data described down in FL this winter before the sales season begins in March 2012.

  2. Sometimes I struggle to put this stuff into terms that make sense to everyone.

    Let’s try this: In every race, most good horses hearts beat 230 times per minute, or nearly 4 times per second. Every time that heart beats, the horse travels a certain distance. If you have 2 horses entered in a match race; one that travels 11 feet per heart beat vs another that travels 12 feet, the former will lose to the latter every single time – even if they appear visually to be equals, as the second horse accomplishes more work via aerobic metabolism with oxygen, and delays the onset of fatigue.

    This can be extrapolated backwards to 2yo in training, and partially explains why a 10sec worker may later lose a 6F race to an 11sec worker, despite bringing a much higher hammer price.

  3. Bill,

    This is a fantastic article and confirms a lot of conceptual thoughts I’ve had. I’m curious how you obtain the numbers to make your calculations though. For example, how do you get the maximum HR numbers?


    • Hi Charlie, nice to have you – thanks for reading. One thing to understand: horses are animals of prey, humans are predators. While it takes up to 90sec of all out exercise for a human to reach a true max HR number, horses hit theirs in as little as 7 seconds. I simply get my numbers during a piece of quick work where they hit 15sec/furlong pace early at age 2. Many physiologists/vets use a protocol on a treadmill with many steps, etc. – but I don’t find that necessary in the field for my purposes.

  4. A scientifically analytic approach (yours is an excellent example) should be the standard in assessing the racing value of a horse ; sadly, there are so many fools who are dying to part with their money. They would rather rely on the easy astrological voodoo of “pedigree analysis”, rather than on the empirically demonstrable outcome of a given pedigree. In addition, the statistical “law” of regression shows that offspring of great (on the racetrack) sires and dams are rarely even nearly as great.
    Such increasing subjectivity, along with out of control hyperbole, has led to an artificially inflated “diamond-like” market among breeders in which “mediocrity” is bred and produced as if it were rarefied excellence. Did you catch the sales price on Royal Delta? $8.1 million? I’d love to see the moneyball stats that were used in assessing her value; I guess even the buyers/suckers correctly calculated that there will be an exponential number of future suckers who will buy her overpriced foals, just throw in a “great” sire like Super Saver or Drosselmeyer.

    I really appreciate the reasoning in your posts. If you are not already, you should be a professor of equine husbandry at some prestigious university.

    • Hello John-

      I certainly understand that most buyers don’t consider ROI when buying horseflesh, it’s more of a hobby for them – but still, -70% ROI at the auctions? We must be able to do better when integrating an outsiders approach consisting of sound principles of equine exercise science.

      Top breeders often privately confide to me that as soon as their sires get hot with a few crops of 10% stake winners and the money flows in – is precisely the time numbers start to regress to the avg of 3% SWs. Just like the stock market: as soon as you see on CNBC that it’s time to buy a stock – the insiders have already made a killing.


    Another good article, but there are ways to analyze a horse from a statistical perspective without spending thousands on heart scans and stride analysis. There are very simple, inexpensive – if not entirely free – methods to finding sound purchases. We at Post Parade Bloodstock employ these methods: this year, we’ve spent $97.5k on 3 2yo’s and have 1 winner from 2 starters, with the 3rd a few weeks away; over $56k in earnings; we also bought 4 yearlings at Keeneland for $82k. We specialize in buying horses for a reasonable rate, and we look for those horses that can win early but will mature into more talented older runners (that’s where the money is). There are many ways to guess at a horse’s talent ~ be careful to not give folks just the expensive options.

    • I am familiar with your methods at “The team sifts through thousands of data points at each auction – pedigree, works, conformation, consignor, medication, and more.” I still believe, and I may be wrong, that you look at info available to anyone – you just may very well be better at deciphering what it means.

      My article pointed out the problems with heart scans and gait analysis, I use neither.

      My methods are not guessing at anything – I only quantify performance based on phsyiological feedback from inside the horse itself. But, I require personal access to the exercising horse in question, which is not easy to get, as you are well aware. I have years of experience with older horses, but this will be my first winter in FL with the youngsters. Initially I doubt my findings will be available to the buying public, I think the data may serve better as a guide for the owners/consignors with respect to reserve prices and deciding when to race, rather than sell, in this market.

      If your overall ROI bests -70% over the next few years, then you are certainly at the top of your game, and I commend your efforts.

  6. Bill,
    Just wanted to say a big thank you for you articles, i have found the scientific aspects of it very interesting and helpful. Im a trainer from rural Australia, only having 2 to 4 horses in work at a time and ride them all myself in track work. I have a heart rate monitor that is about 7 years old, and am very interested in upgrading to a newer version and would like to know your opinion on what would be the best one to buy for someone in my situation.
    Thank You,

    • Hi Doug, thanks for the note.

      Sounds like you have a great set up there in Australia.
      The newest HR monitors incorporate GPS and extensive software analysis, anyone who buys from me gets lifetime email support.
      Here is the link to my online store for the Polar Equine RS800CS G3, please let me know if you have questions-

  7. Bill,

    If we are comparing individuals, how do we compensate for variables like conditioning? When we look at a horse for sale, whether a racehorse or, say, a 3-day eventer (where I come from), we cannot know for sure why the horse’s number’s do match or don’t match up with numbers from an ideal specimen.

    I’m just wondering how we can compensate for these variables in a scientific way.


    • Great question Jessica-

      The answer is: you cannot do so. So we attempt to gauge all the horses we can get to within the same week of training. For instance, all the horses tested along with Battier were done within the same week – after 5 months of training at a specific facility. All were headed for the sales ring, so all were roughly the same spot along.

      Now in my daily work assessing HR/GPS in racing thoroughbreds, I have to rely on the trainer’s info as to how ‘fit’ one should be. On the youngsters at the farm, we factor in which ones got a later start in our comparisons. Comparing horse to horse is not near as valid as comparing a horse to himself over months of time, of course.

      • Endurance and 3-day riders have side awards for the Best Conditioned Horse, and I think they base it on recovery time. Maybe TB Racing should have that! It would at least tell us which trainers are putting in the sweat equity!

        In this day and age, with heart monitors and such, maybe all horses should be tested after racing by a vet or vet tech. Or better yet, after each publishable breeze. That way fitness for duty could be determined. People should watch XC day in a 4-star event or steeplechases like the Grand National to see how fatigue manifests itself. Jumping for diminishes sharply. It is then easy to imagine what a tiny mistep at speed can result in.

  1. Pingback: Thoroughbred ‘Moneyball” Unearths a Gem at 2012 FT Timonium Sale | ThoroEdge Equine Performance

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: