Aug 23, 2012

Fit vs. Fragile


Rail Trip
A few years ago, I wrote about the Freshened Horse Fallacy, the theory that the modern-day thoroughbred simply cannot withstand the rigors of day-to-day training and — heaven forbid — racing on a consistent basis.

I pointed out that much of this hysteria is based on the belief that thoroughbreds are getting weaker and more fragile due to the emphasis American breeders are putting on speed, rather than stamina. 

The premature retirements of both I’ll Have Another and Bodemeister, the 1-2 finishers in this year’s Kentucky Derby and Preakness Stakes, seems to back this contention up, as does the recent tendon tear that unethically targeted last year’s juvenile champion Hansen and likely relegated his career to the history books as well.

That horses are making fewer and fewer starts is indisputable. In 1975, the average thoroughbred went to post 10.2 times a year; in 2011, the average number of starts per year was down to 6.2.

Of course, whether or not this has to do with dubious DNA or a changing economic and social landscape is very much in doubt. After all, does anybody seriously think that we’ll ever see another Triple Crown winner compete past the age of five, which both Assault and Citation did (in the 1940s)?

What’s more, if American stud farms are, in fact, breeding for speed these days, they’re having about as much success as Kris Jenner did breeding for class and dignity. While all eight of the quarter horse world records for distances up to 550 yards were set in the last three years and standardbred records seem to fall every week, only a handful of thoroughbred world-record times have been recorded in the new millennium. Some, like the seven-furlong mark, date back to when the Internet was merely a sparkle in Al Gore’s eyes.

Personally, I think the notion that breeding for speed has led to the decline of the thoroughbred racehorse is largely bunk. Likewise, the view that more rest and fewer races is the only way to effectively deal with this “delicate” situation is also malarkey.

When I penned “The Freshened Horse Fallacy,” I presented some test data to prove this and I want to do the same thing here, only with new, up-to-date statistics. But before I reveal the numbers, let me first establish the ground rules, as set in that original piece:
Of course, the biggest challenge one faces when attempting to prove or disprove a racetrack “fact” is obtaining truly independent variables. The reality is that almost no single factor contributing to the outcome of a horse race can be easily isolated. Take, for example, speed and form — what’s the real difference? Typically, a horse that runs fast also runs well, right? After all, it’s not often that a 30-length loser will post an outstanding Beyer figure.

So, my first hurdle was distinguishing between a “freshening” and layoffs resulting from injury or infirmity. Thus, I decided to concentrate solely on post-time favorites (ignoring entries). That way, I could be reasonably certain that I was apprising only those contestants that had shown at least a semblance of class and form in the recent past.

Now, does this ensure an autonomous sample? Of course not. Obviously, the date of a horse’s most recent outing is going to influence the crowd’s betting habits, but at least it helps eliminate those hapless nags that neither racing nor resting will aid.

First, I looked at favorites as a whole (provided they were single betting interests with at least one lifetime start) from assorted races run during 2004-2009:

Races: 5,786
Won: 2,075
Rate: 35.9%
ROI: -15.49%

As you can see, these figures are right in line with long-term national averages. Thus, my database would appear to be “fair.” Next, I looked at favorites that were coming back on less than 10 days rest:

Races: 250
Won: 98
Rate: 39.2%
ROI: -8.36%
At this point, let’s take a DeLorean back to the future and take a peek at the current figures, culled from my brand new database consisting of 4,873 races. First, all the favorites:

Races: 4,873
Won: 1,811
Rate: 37.2%
ROI: -16.82%

What immediately strikes me, of course, is that although favorites are winning more now than before (37.2 percent vs. 35.9 percent), the ROI is even worse (-16.82 percent compared to -15.49 percent). Just more proof that value betting is the only way to profit in today’s game.

However, that’s not what my test and this article are about. So, with that in mind, let’s see if the current crop of thoroughbreds can stand up to the “rigors” of a recent race. Below are the digits for favorites competing within 10 days of their last start:

Races: 225
Won: 97
Rate: 43.1%
ROI: -10.11%

Once again, we see a huge improvement in the stats — an improvement that seems to argue against an increasingly brittle breed. This test also demonstrates why one should (at the very least) be hesitant to accept horseracing “truths” at face value.

Far too often, they are anything but.

Six-Figure Saturday & Sunday

A couple of $1 million Grade I events highlight this weekend’s racing action. On Saturday, the top (remaining) three-year-olds do battle in the Travers Stakes at Saratoga; then, on Sunday, the best older horses on the West Coast square off in the Pacific Classic at Del Mar.
 
In the former, I’m intrigued by the ever-improving Street Life, who just captured the Curlin by 1 ¾ lengths on July 27, as well as Liaison and Neck ‘N Neck, both of whom had wide trips vs. probable race favorite Alpha in the Grade II Jim Dandy on July 28.

Where the surf meets the turf, I’ll be keeping my eye on Rail Trip, who recorded a race-best +6 late speed ration (LSR) in his last race, the Grade II San Diego Handicap. The old guy still has some spring left in his legs and he should get a great... well, rail trip... in Sunday’s feature.

On the other hand, I’m completely tossing Dullahan, who is starting to remind me of Ice Box. Yeah, his non-effort in the Haskell can probably be excused — I’m not convinced the son of Even The Score cares for the dirt, much less Monmouth Park’s speed-favoring surface — but, honestly, this guy’s resume hardly screams greatness. And given that the Classic will mark his first try against older foes, I’ll take a pass.

For FREE Brisnet past performance that include my speed rations (ESRs and LSRs), check out the links below:

39 comments:

  1. As if on cue, Willy Beamin wins the Grade I King's Bishop Stakes (at 11-1) on three days rest.

    ReplyDelete
  2. That's some leap of logic to conclude from that "study" that horses are no less brittle today. Rather than proving your thesis it's more likely you uncovered the following facts: Recency is important. That's a given in the one group. a complete unknown in the other. Perhaps in the large sample you should have eliminated those favorites that hadn't run with 45 days. Secondly, it may be that favorites running over the sme track outperform the larger sample. And third, bettors as a whole may have some aversion to betting quick turnarounds. As for the seeming paradox of an increased win percentage but a lower ROI, what you have there is decreasing field size. Some of the others mistakes and mischaracterizations you make include the term "breeding for speed" when you mean "brilliance" and the assertion that speed and form are somehow synonomous or uncouplable. They are completely different concepts. Form is trajectory, recency and relative performance. Speed as measured and defined by you as a horse's fig is dependent on conditions and dynamics. Form comes and goes in a cycle. It involves a series of races. Speed is a single race, final time measure dependent on all those things that make up "trip."

    ReplyDelete
  3. My first sentence should, of course, read "no more brittle."

    ReplyDelete
  4. Not sure you're numbers prove it, but you are right. That theory is complete nonsense.

    ReplyDelete
  5. This comment has been removed by the author.

    ReplyDelete
  6. Where to start, Kyle:

    1) “That's some leap of logic to conclude from that 'study' that horses are no less brittle today.”

    I am not asserting that coming back quickly necessarily means the breed is less brittle, but the argument put forth by many horsemen and fans is that horses need more rest these days BECAUSE they are more fragile. My statistics suggest that at least that part of the argument is bogus.

    I have long maintained that the rapid rate of evolution (or should I say de-evolution?) that would need to take place for the breed to become more fragile is what I find most disconcerting.

    2) “Rather than proving your thesis it's more likely you uncovered the following facts: Recency is important.”

    Not according to those who preach about the fragility of the breed. To them, racing is the enemy; hence, the calls to re-structure the Triple Crown series to give the horses more time to "recover."

    3) “Perhaps in the large sample you should have eliminated those favorites that hadn't run with 45 days.”

    Again, the argument is that horses can no longer stand racing consistently. Why would I limit the study to horses that HAD raced recently?

    To be continued...

    ReplyDelete
  7. ... continued.

    4) “Secondly, it may be that favorites running over the same track outperform the larger sample.”

    The reason I used favorites is because, for the most part, there is very little variance in the stats regardless of what criteria you use to parse it. Yes, there’s a SLIGHT advantage to racing over the same track, but not the 6-7 percent you see in the data I presented.

    5) “And third, bettors as a whole may have some aversion to betting quick turnarounds.”

    If this were true, the ROI would be worse. Given the increased win rate, it is clear that the prices on horses coming back within 10 days are LESS than those on steeds that are not (some of this might be explained by field size too).

    Not sure what your point here is. Mine was/is that, despite what one hears about the benefits of layoffs and the difficulty of winning on short rest, the stats don’t bear it out.

    6) “As for the seeming paradox of an increased win percentage but a lower ROI, what you have there is decreasing field size.”

    This one really stumps me. Please show me the math that bears your contention out. Yes, field sizes have dropped, but that’s got nothing to do with ROI directly — it affects the win rate, which BY EXTENSION affects the ROI.

    In fact, you might be surprised to learn that smaller fields (at least in my sample) actually produced a higher ROI, relatively speaking, than larger ones.

    7) “Some of the others mistakes and mischaracterizations you make include the term ‘breeding for speed’ when you mean ‘brilliance.’”

    True, industry types often use the word “brilliance” when discussing speed-laden pedigrees, but the “breeding for speed” mantra has been used by many and, frankly, WHATEVER term one chooses to use is irrelevant.

    (“Brilliance” is generally not used as a synonym for speed OUTSIDE of racing circles.)

    8) “The assertion that speed and form are somehow synonomous or uncouplable. They are completely different concepts. Form is trajectory, recency and relative performance. Speed as measured and defined by you as a horse's fig is dependent on conditions and dynamics. Form comes and goes in a cycle. It involves a series of races. Speed is a single race, final time measure dependent on all those things that make up ‘trip.’”

    I’m afraid this is probably a semantics argument. However, I stand by my assertion that, generally, if one says a horse is “in good form” one is not referring to a horse that ran significantly slower than its rivals last time (barring a reason for that slowness).

    OF COURSE, form is relative — so is speed. This is why there are such things as par times. What’s more speed, like form, is NOT an absolute as you seem to imply. Speed figures also vary, often greatly — sometimes due to conditions, sometimes due to trip and sometimes because the horse just doesn’t feel like running that day.

    I’ll offer you a challenge: Show me a statistical model that completely and effectively separates speed from form and I will believe you that the two are totally unrelated.

    I don't think you'll be able to.

    ReplyDelete
  8. The numbers don't really PROVE that the breed is more/less brittle (as I noted above), anonymous; they simply DISPROVE the notion that today's thoroughbred needs more time between races to perform at its best.

    ReplyDelete
  9. I guess I have my work cut out for me. Let's start with our one area of agreement: genetics seems to suggest 30 or 40 years is too short a time to have a significant impact on the thoroughbred breed. However, it would probably say the same about the species "MLB Pitcher" and like our favorited four-legged athlete that one is also showing unmistakable signs of a rapid lessening of durabilty over a similar time-frame. So, whether it be genetics, environment (and all the factors that make up that term) or a combination of the two these things do happen.
    To argue that horses are no less durable today is to argue against the self-evident for just about handicapper whose experience spans the transitional period of the last 30 years. You gave some of the pertinent stats. You mentioned this year's Triple Crown experience. Two things that don't happen anymore, or happen very rarely, that used to happen with regularity - Two year old stars going on to become standout three year olds with careers that extend beyond June of their sophomore years and Triple Crown stars coming out of the series to mature into Breeders Cup Classic contenders that fall and star older horses the following season. It is not the long, lucrative reach of the breeding shed plucking sound animals from the racetrack, but by and large, one infimity or another forcing them into pre-mature retirement. Of course it goes beyond the Triple Ctown. One for instance, it used to be commonplace for fillies to contest Saratoga's Test Stakes the first Thursday of the meet and come back 10 days later in The Alabama. And after that, continue on racing every couple of weeks through the fall. Now, that kind of thing is so anachronistic that those two races have been completely de-coupled. So, are horses just as stout? Trainers certainly don't think so. I think the only debate is why, not if.

    ReplyDelete
  10. Now to the other points: The continuing indisputable benefit of recency in no way argues against a less durable animal. Runners still require fitness for optimum performance. Where fragility enters the picture is how long that performance level can be maintained.
    Running back over the same track wouldn't account for the entire difference you found. It's the combination of factors I cite that would.
    As to the aversion to betting quick turnarounds on which I speculated, You seem confused. The less bet on a certain set of winners the higher the ROI. Now, I must admit,I was looking at your first set of stats which showed only a 3.3% increase in winners but a jump of more than 7% in ROI. But even your newest numbers show a slightly increased ROI relative to win percentage.

    ReplyDelete
  11. Oh, the smaller field size thing. The math? The smaller the field size the higher the the percentage of winning favorites (that's a mathematical certainty over a large enough sample) and the greater the the effect of takeout; hence, lower ROI.

    ReplyDelete
  12. To begin with, Kyle, the pitcher example doesn't work for me. We live in an era of specialization and sports is no different. Comparing guys like Cy Young to modern-day pitching stars is akin to comparing Sammy Baugh to Tom Brady. Players are bigger, stronger and faster these days — and that comes at a price (the concussion issue in the NFL being one prime example of how pro sports have changed).

    However, there is a statistical lesson to be learned from baseball. Just as you note that horses running less often today is self-evident, so is the fact that today’s batting champs do not have the averages of years gone by.

    From 1911-1941, the average MLB batting leader maintained a .383 average; from 1981-2011, the average was .350.

    Now, in the racing world, we would immediately conclude that players are being bred improperly and weep and gnash our teeth over this fact (just kidding… kind of). But is it fair to say that today’s hitters are inferior to those of yesteryear? Again, the argument seems to turn Darwin’s theory on its ear.

    In fact, it is hardly a coincidence to me that Ted Williams was the last player to lead the league in hitting with a .400 or better average in 1941… a mere six years later, Jackie Robinson broke the color barrier and that led to a vast expansion of the MLB talent pool.

    Similarly, the RIDICULOUS valuations of thoroughbreds that began in the 1970s and 1980s has led, I believe, to a “kid’s gloves” approach to training that has left many racehorses underprepared to compete on a regular basis.

    Citation ran through numerous injuries (many of which would not have been detected like they are today) because his owners wanted him to become the first millionaire racehorse. Citation met that goal, retiring in 1951 with $1,085,760 in earnings from 45 starts and 32 wins. That is the equivalent of about $9,567,383 today — an amount that Curlin exceeded in 16 starts and 11 wins.

    What’s more, Triple Crown entrants specifically are running far less often now than even their compatriots, suggesting again that their value off the track is considered to be greater than their value on it.

    A weakening of the breed? No. A weakening of the sport? Most definitely.

    ReplyDelete
  13. "As to the aversion to betting quick turnarounds on which I speculated, You seem confused. The less bet on a certain set of winners the higher the ROI."

    The first data set:

    37.2% winners and a -16.82% ROI = $4.47 average payoff.

    The second data set:

    43.1% winners and a -10.11% ROI = $4.17 average payoff.

    As you can see, the "quick turnaround horses" (second data set) are getting bet heavier, thus invalidating your point (minus the effect of field size, which I already offered as a caveat).

    ReplyDelete
  14. "The smaller the field size the higher the the percentage of winning favorites (that's a mathematical certainty over a large enough sample) and the greater the the effect of takeout; hence, lower ROI."

    This is just factually wrong in regard to the takeout. Takeout is a percentage of the pool and has nothing to do with field size, outside of the fact that larger fields TYPICALLY generate greater pools.

    ReplyDelete
  15. I don't want to get into the baseball thing too much. But, A) It would not be unfair to say hitters are inferior today in some ways, especially when it comes to making contact. And B) In baseball there are clear reasons. Night games, expansion, and relief pitching.
    On our central topic, I'll sum up this way. I think it's sophistry to suggest horses are just as stout today as yesteryear. But I'm not saying breeding alone is the culprit. It may not even be central. How foals are raised, how they are prepared for sale, financial considerations of all kinds and trainer methodology are more likely top be the culprit. It's the end result I'm concerned with. Personally, I think pharmacology is at the root of the problem. I think it has enabled trainers to train the way they do, or maybe forces them to. There's a feedback loop there. It's distortive. And it does end up extending to the breeding shed. The fact that horses can and do win on short rest doesn't disprove a thing. It doesn't even suggest that those who see a weakening of the breed are wrong. Willy Beamin, by the way, is now going to the sidelines for a freshening. He was off for two months from late June to late August and now he'll get at least another two months rest.

    ReplyDelete
  16. Average mutual doesn't matter. ROI is the only thing that matters. That's as much as I can elaborate on that. Either you understand that or you don't. As to takeout and field size: The smaller the field size the less volitility, the greater the market efficiency, the greater the effect of takeout over time. That's why your numbers show favorites winning at a higher percentage but returning less overall.

    ReplyDelete
  17. Maybe this is where you're confused. You're citing the difference between the two recent samples. The influence of field size is seen in the differences between the earlier study and this more recnet one.

    ReplyDelete
  18. Derek, ask yourself this. Which would I rather bet into: a five horse field with a 15% rake or a 10 horse field with an 18% take? Do you play poker Derek? This is along the same lines as when Hold 'Em players ask themselves, "over the long haul, what is my expectaion playing these two pocket cards?"

    ReplyDelete
  19. "Average mutual doesn't matter. ROI is the only thing that matters. That's as much as I can elaborate on that. Either you understand that or you don't."

    Kyle, you continue to display an arrogant attitude ("you either understand that or you don't") despite being dead wrong on this one.

    ROI is a function of both success rate and the average mutuel: Win% x average mutuel / 2 - 1.

    Obviously, there are other ways of computing it, but to say that the average mutuel, which is just another way of expressing the odds, doesn't matter is... well, wrong.

    ReplyDelete
  20. "Which would I rather bet into: a five horse field with a 15% rake or a 10 horse field with an 18% take?"

    Clearly the former, Kyle, but that is NOT the way racing works. Where did you hear that the takeout varies based on field size?

    ReplyDelete
  21. Poker bores me (LOL). I can't get into it.

    ReplyDelete
  22. As for your "first" response (August 28, 2012 7:32 AM): It is far different to claim that breeding for speed ("brilliance" if you prefer) has significantly altered/weakened the breed and "But I'm not saying breeding alone is the culprit. It may not even be central."

    If you had led with that statement instead of trying to educate me on points I don't think you yourself fully understand (takeout, ROI, efficient markets, etc.), we'd both be living productive lives instead of wasting our time arguing on the Internet (LOL).

    That said, I've enjoyed the debate.

    ReplyDelete
  23. I tried, so I don't consider it a waste of time. Maybe others have benefitted. One last time on ROI. The more bet collectively on a set of winners the lower the return. The less, the higher. How often horses of either set win is beside the point.
    Yes, takeout is not currently tied to field size. It should be. But racing does work that way - field size overcomes take at a certain point. You admit that. You just don't believe it, I guess. Since you like studies, here's one I'd suggest. Take a set of races with 6 or fewer horses from a low takeout track(s) randomly selecting a runner. Record the return. Do the same at a higher takeout track with races containing 9 or more betting interests. The results I think would be interesting.

    ReplyDelete
  24. And by the way, I know what I'm arguing is a paradox. Isn't it also paradoxical that betting every horse to win over the history of the Breeders Cup would result in a profit? Why isn't there a loss of 17 or 18 percent?

    ReplyDelete
  25. You just refuse to admit when you're wrong, Kyle :).

    "The more bet collectively on a set of winners the lower the return."

    Yes, we agree! The amount bet on each horse, minus the takeout, is how the odds are computed. However, notice that you specify the amount bet on the WINNERS. This is an important distinction and it is why the winning rate matters when computing ROI.

    The only reason the second sample yielded a better ROI is because there was a higher winning rate, which overcame the slightly lower mutuel.

    But, neither circumstance (higher winning rate, better ROI) makes your point that "bettors as a whole may have some aversion to betting quick turnarounds."



    ReplyDelete
  26. Yes, takeout is not currently tied to field size. It should be. But racing does work that way - field size overcomes take at a certain point."

    I can't make heads or tails of this statement. What do you think your mythical study would show? Certainly you're not suggesting that smaller/larger fields completely eliminate the take at some point ("field size overcomes take at a certain point.")?

    I COULD entertain the idea that smaller fields might lead to a more efficient market (as you seem to have suggested previously), but it will NOT eliminate the take altogether.

    ReplyDelete
  27. "And by the way, I know what I'm arguing is a paradox. Isn't it also paradoxical that betting every horse to win over the history of the Breeders Cup would result in a profit? Why isn't there a loss of 17 or 18 percent?"

    Kyle, what you're arguing is a lack of understanding pertaining to math and statistics. I saw many of these same type of arguments when I opined on post position and the KY Derby.

    First of all, random bets on a typical race would not result in a 17-18 percent loss; instead, the number would be in excess of 20% due to what many scholars have termed the "favorite-longshot bias."

    In simple terms, the 17-18% loss would only exist: a) Minus breakage; b) If every horse had an equal chance of winning.

    Neither a & b are true a typical race.

    ReplyDelete
  28. Listen, we're talking relatives, Derek. And I thought a common gambling experience allowed some short hand. But if you're going to be obtuse....OK, to be more exact. There is a point at which field size more than outweighs some takeout difference for the competent gambler as reflected in his bottom line. As for the study I suggested, what I think you might see are ROIs that are closer to each other than the difference in takeout would suggest.

    ReplyDelete
  29. I do bet the races. Would you like to involve yourself in a public contest where I could show you how much more knowledgable I am than you?

    ReplyDelete
  30. Kyle, the problem is you're making blanket statements, which are factually wrong, and, then, you're get increasingly more arrogant when I disagree with you.

    What is obtuse is not understanding that math is not a social science -- things don't compute a certain way just because you want them to.

    Sorry, but a competent gambler can win in a 5-horse field, a 9-horse field or a 20-horse field.

    Yes, one can argue that takeout has a more significant REAL effect when one wins more often, but that is akin to saying that one pays more taxes if one makes a greater income. (Not to mention that higher prices generally mean a lower win percentage, higher volatility and greater risk.)

    ReplyDelete
  31. "I do bet the races. Would you like to involve yourself in a public contest where I could show you how much more knowledgable I am than you?"

    In your own mind, I'm pretty sure you're more knowledgeable than anybody, Kyle.

    As for your challenge, I no longer make my bets public on a consistent basis, but I'd be HAPPY to post your selections and track your results.

    Based on some of the things you've said, if you even make a 5-10% profit, I'd be impressed.

    ReplyDelete
  32. Write me on Facebook and we'll set it up. This should be fun!

    https://www.facebook.com/Youbet.DerekSimon

    ReplyDelete
  33. First off, what blanket statements have I made besides the ROI thing? And really, the genesis of our arguement - the possible aversion of the masses to bet quick turnarounds - is a pretty minor point. And as to gamblers being able to be successful over the long run in five horse fields - I'm unfamiliar with such an animal.

    ReplyDelete
  34. None of my techniques takes field size into consideration. Using fair odds eliminates the need for that.

    Back to this challenge, though: Write me on Facebook, I've got a great idea for you. (If you're not on Facebook, it's easy to sign up.)

    ReplyDelete
  35. Yes, value is what matters be it a 3 horse field or a 12 horse field. The shorter the field, though, the less often you find value in my experience.

    ReplyDelete
  36. An interesting take on the 'Fit vs. Fragile' debate in horse racing. The stats provide food for thought. Looking forward to more insightful analyses from you, Derek Simon!
    File Divorce in New York city

    ReplyDelete
  37. There are no words to describe how good poppy playtime chapter 3 is. To skip it would be to miss it the most.

    ReplyDelete