James Willoughby
One of the roles of the analyst, particularly in the sporting arena, is to separate luck from skill when observing positive outcomes. It is highly possible to be an ordinary trainer with a lot of good horses and convince the market that you have skill, simply because horses make humans look good, and rarely the other way round.
— James Willoughby / TDN Journalist

Their exploits at Cardiff's Millennium stadium against France last evening, wasn't the only thing the Irish got right on the weekend. Yet the fellow we're talking about has been doing it week in, week out, year in, and year out for decades now. Just Friday we penned a piece on the hit rate of a successful team ("Who's Who in the Thoroughbred Zoo") and James Willoughby has reciprocated with an equine version of what a good "driver" can do for a good machine, a´la Lewis Hamilton and Mercedes.

An Aidan O'Brien post-race paean is often greeted with a degree of suspicion nowadays, but so brilliant is his G1 Dubai Dewhurst Stakes winner Air Force Blue (War Front) that the trainer's "He's something that we haven't had before" in the Newmarket winners' enclosure seemed phlegmatic. But O'Brien's comment surely could be applied reflectively to the trainer himself. One of the roles of the analyst, particularly in the sporting arena, is to separate luck from skill when observing positive outcomes. It is highly possible to be an ordinary trainer with a lot of good horses and convince the market that you have skill, simply because horses make humans look good, and rarely the other way round. O'Brien has done it so many times and put such an obvious stamp on the brilliant horses that he nurtures, that a statistician has long had to reject the null hypothesis that he is just another trainer favourably situated. With Air Force Blue standing there  being cooled off, it's hard to remember that the careers of elite racehorses are highly path-dependent, and so many wrong turns are possible along the way that it actually is possible to have near-unlimited funds and still not dominate the sport.

Unless you have O'Brien, that is. The source material, from which a billion outcomes are possible, was paraded during the week of the Dewhurst as the offerings at the Tattersalls Book One sale. The prices paid for prospects, like the 2.1m-guineas Dubawi (Dubai Millennium) filly acquired by Coolmore, seem irrational to the itinerant observer, with some writers tempted to sneer by citing expensive failures of the past from Snaafi Dancer (Northern Dancer) to Hydrogen (Galileo) to The Green Monkey (Forestry). But this cherry picking should not pass for sage analysis; instead, a wider sweep of sales data proves that prices paid bear a highly statistically significant relationship with outcomes on the track. It is just that all similar relationships involving prediction or projection involve noise.

It is often said that buyers at Book One are chasing a dream or playing a high-stakes form of Lotto, or something. But it's better to frame their incentives in economic parlance: they are attempting to get to the elusive right-tail of the distribution of all racehorse talent, and to do so involves taking on considerable risk. Of course, this is no attempt to disguise the economic reality of the sales ring; plenty of transactions fall outside the bounds of rationality, there are bubbles like the present one surrounding the progeny of Dark Angel (Acclamation), and the aggregate of all spend has a negative expectation. But, back to the concept of path-dependency, investing large sums in bloodstock can go wrong at several stages. Whoever is buying your horses can only exert so much effect beyond the relationship between price and realized ability; as in the markets, favourable results in any period regress to the mean as the sample-size of purchases increases. But the identity of the individual training the horses is many times more important.

Imagine if a bookmaker offered 10-11 (-110 as they might quote you in Las Vegas) about an under-over on the career-best rating of the Dubawi filly sold to Coolmore. Even with O'Brien as her trainer, the data suggests it is not possible to go much beyond Timeform 100. There are just too many negative outcomes inherent in the development of her career. But, what might be the revised Timeform rating if a trainer were selected at random from all those with more than 50 horses? What would be the under-over then? This question - what is the marginal effect of O'Brien as a trainer above replacement level - is the one that other sports franchises realized they need to ask about their human talent. Amazingly, it seems like racing still relies on counting stats like winners, prize-money or championships in naive belief in the cream-rises-to-the-top effect. Yet, in the world around it is clear that we are all a function of our opportunity and our capabilities, with the ratio between them nebulous but surely with a much greater weight on the former than the latter.

Certainly, there are trainers like O'Brien and jockeys like Ryan Moore whose marginal effect is both real and positive, but to get to the root of who is really good, it is necessary to first discount opportunity. Any operation which can't or won't bring itself to ruthless self- determination in this manner now has to bear the consequences in an increasingly competitive world within and without racing. When O'Brien first got the job at Ballydoyle in 1996, he could have turned out to be far less the trainer we know now, even though it was possible to form a strong, prior belief in his capabilities from his former exploits. But, as the sample-size has expanded, it has become less and less likely that we have been fooled by randomness or been guilty of not separating his skills from the lavish context of Ballydoyle and its Coolmore-driven equine resources. And, as for the idea of ever letting him walk away from your operation, you would have to be out of your mind.

Extract from Thoroughbred Daily News by James Willoughby
Aidan O'Brien & Camelot / Coolmore (p)

Comment