When the Festival concludes next Friday, praise will be variously doled to the talent involved, writes Tony Keenan. Horses, trainers, jockeys, maybe even stable staff, will get credit for their efforts in victory. Much of it will be deserved but the one thing unlikely to be mentioned is luck.
Part of this is simply our thinking biases; humans operate under the illusion of control, overestimating the role we play in outcomes. Another aspect is that luck is hard to quantify in racing; we can all recall specific examples of luck in action, when a horse fell when seemingly going best or failed to get a clear run when travelling strongly or simply a narrow defeat, but gauging trainers who are the victims of variance over time is more difficult. Surely not all of them are equally lucky, especially at the Cheltenham Festival where there are only 28 races, a very small sample size.
Data analysts or sabremetricians have sought to quantify this in other sports, specifically those based in the US. Pythagorean expectation, the formula that estimates how many games a team should have won based on their scoring, have proven a better predictor of future success than past win-loss records in sports in baseball, basketball and American football. These theories have crossed into European soccer too with numbers on shots, shot quality and expected goals now playing a part in some sensible conversations on the sport.
Translating this into racing isn’t easy but it seemed worth a try going back as far as the 2010 Festival.
Rather than taking just one criterion, I decided to use three to see if the same trainers were unlucky across the different metrics. Firstly, the old favourite expected winners -the number of winners a trainer should have had judged on market prices - to see who was lucky and unlucky, overachieving and underachieving. From there, I took the number of seconds and placed runs relative to winners to uncover who was getting close without winning.
Finally, I looked at the in-running markets from Betfair for all the races since 2010 to see how many odds-on in-running trades trainers had, as sometimes the place results may not tell the whole truth, for instance when a horse that looked set to be involved in the finish fell close home. I used 2.0 as my cut-off point as an odds-on trade reflects a view held by someone (rightly or wrongly) that a horse was more likely than not to win a race at a given point.
Cheltenham Festival: Trainer Performance Based on Market Expectation
|Trainer||Actual Wins||Expected Wins||Difference||Actual/Expected|
|H. De Bromhead||3||2.4||+0.6||1.25|
It seems scarcely credible but these figures suggest the Festival markets still hasn’t totally caught up with Willie Mullins; he is outperforming expectations despite breaking records at the meeting.
Perhaps this year, when the yard has had so much bad luck ahead of the meeting, will finally see his runners overbet. Paul Nicholls could be Mullins of five years in the future; after a period of being top trainer at the meeting (he won it five times between 2004 and 2009), he now has one of the poorer records among the top trainers, with only Alan King having a lower actual/expected figure.
This is the top group of trainers in terms of winners sent out at the meeting, however, and unsurprisingly most are doing better and/or are luckier than the betting suggests. That could well simply reflect their skill and the quality of their horses but one obvious conclusion is that there must be an awful lot of smaller yards really struggling for a winner who have negative figures.
Gordon Elliott and Jonjo O’Neill are two that stand out in terms of luck though with Elliott it seems likely the market will take full cognisance of the level he is currently operating at; whereas in past seasons, he was slightly under-the-radar, now he is a presumptive Champion Trainer with the favourite or second favourite in seemingly every handicap at the meeting. O’Neill is a different case and his results might be down to how his stable performs through the winter; it seems that every March, his runners come into the Festival under a cloud and the markets have to have them at bigger prices as a result.
Alan King is one of the unluckiest big trainers – a point we’ll return to later – while Jim Culloty is the luckiest and it’s not even close. His actual over expected ratio is off the charts but this looks a case of pure randomness rather than skill; everything else we have seen in his training career thus far says he is not this good and, realistically, no trainer could maintain such figures. Trusting those figures and betting his horses at the Festival would be to fall prey to an extreme form of survivorship bias.
Cheltenham Festival: Seconds and Places
|Trainer||Wins||Seconds||Difference||Places (2nd, 3rd and 4th)||Winners to Places Ratio|
|H. De Bromhead||3||4||-1||10||3.33|
In terms of simple winners to seconds difference, Mullins comes off best again. Philip Hobbs is next in with five more winners than runners-up while Rebecca Curtis could well be called "the milk-woman" in that she always delivers with not a single runner-up and only three places to go against her four winners. The unlucky trainers in this regard are Paul Nicholls, Mouse Morris, Martin Keighley and Tom George.
Winners to place ratio is simply places divided by winners; the places here don’t include winners. By my reckoning, a ratio of above 3.00 suggests bad luck while below suggests good luck; there are 3 places available in each race with only one win. Alan King’s misfortune is the one that jumps out here with an amazing 25 places to four winners for a ratio of 6.25 which is more than double what would typically be expected. Both Mouse Morris and Noel Meade have higher ratios but King’s comes from a bigger sample size. Nick Williams, too, has had a lot of horses run well without winning and is still waiting for a first Festival winner.
Cheltenham Festival: In-running Trades
These in-running histories would surely make for grim reading for many a punter though perhaps not as much as they do for Paul Nicholls; in back-to-back renewals of the Gold Cup in 2010 and 2011 he watched both Kauto Star and Denman trade odds-on in-running before getting beaten. That’s rough.
Nicky Henderson – 2011 Supreme with both Spirit Son and Sprinter Sacre – was only other trainer that happened to in the period covered. These Betfair numbers basically back up a lot of what we’ve seen already: Willie Mullins, Jonjo O’Neill and Nigel Twiston-Davies have been lucky; Tom George, Martin Keighley and Nick Williams have not.
So who should we be looking at for some regression, positive or negative, next week?
Overall, Willie Mullins, Rebecca Curtis and Jonjo O’Neill might see their winners drop while Tom George, Martin Keighley, Noel Meade and Alan King could be heading the other way. That of course depends on whether you think they were lucky or good and as they always say, it’s better to be the former than the latter!
- Tony Keenan