The preceding pieces in this series dealt largely with the theory of handicapping, but with some working examples (which were not cherry-picked to prove my points: believe me!).
It is worth considering some more real-life examples which both prove the validity of the approach overall and identify one or two areas in which that approach needs to be more sophisticated. I have intentionally cut a few corners to this point in order to get the basic message across without too many digressions.
The following are the five main championship events at the recent Cheltenham Festival assessed by means of race standardisation. The lighter grey shaded areas are inputs, including historical BHA ratings and (in the final column) the BHA ratings for this year’s race. The standard assessment, calculated from the available information, is highlighted in red.
Let’s go through the races one at a time.
The historical standards vary between 160 and 169, and average at 164. The BHA’s assessment of Epatante at 162 is a bit below that par figure, but is almost in line with what would be expected if the last four years were used rather than the last five (the 2015 race, won by Faugheen, is something of an outlier).
Due to her sex allowance, Epatante comes out approximately 3 lb worse than runner-up Sharjah but would likely be receiving that allowance again if the two were to meet in the future.
The difference between the BHA’s poundage for margins beaten and the time-based calculation of the same means that fifth-placed Petit Mouchoir is rated only 6 lb worse than Epatante by them but 9 lb worse using time.
There is ample evidence from the next-time performance of winners, seconds, thirds etc, that the BHA’s pounds per second (from which pounds per length should be derived) is lower than it needs to be and fails to give winners and prominent finishers due credit.
The BHA’s assessment looks too low based on the race standardisation figures employed so far, but there are a number of reasons to believe that is at least partly justified.
First and foremost is that, for simplicity’s sake, we have not been using a field-size adjustment so far. Both maths and evidence shows that race strength is higher when field sizes are big, and lower when they are small.
Late defections meant that, with five runners, this year’s Champion Chase had the smallest field since Badsworth Boy beat four rivals in 1985. Those five runners were approximately half the number that had turned up in recent years.
There should be a downwards adjustment when field sizes are smaller than previously and an upwards one when the opposite is true. This is particularly true at the lower end of field sizes.
In addition, experience has shown that it is sensible to have a “cap” on cumulative margins, so that horses like Sceau Royal (beaten a long way into last of five) do not skew the figures disproportionately. Beyond a certain point, running poorly is simply running poorly and the full magnitude of that running poorly is unimportant.
Even after applying these, the standard for this year’s Champion Chase winner is in the mid-170s, which may be too high. You had the case of an unexpectedly depleted and small field, in which the clear form pick (Defi Du Seuil) underperformed. Race standardisation may not be the best solution in such circumstances.
Along the way, I found that the BHA’s site was difficult to use for extracting historical ratings, let alone for calculating what various horses’ “master ratings” would have been going into those races. By using Timeform’s ratings instead, and adjusting them to the BHA level, I got a figure of 173.5 for Politologue using horse-level standardisation.
The race standards agree with the BHA assessment for the winner of this year’s Ryanair Chase, but with lesser ratings for those behind the first three on account of the different poundage allowances employed.
Race standards rate the Stayers’ Hurdle higher than the BHA did, but only just. Both approaches have Lisnagar Oscar improving considerably but still a somewhat substandard winner. This would have been different if he had strung out his rivals by wide margins, but there were only eight lengths between the first six.
In this instance, the BHA ratings are higher than the standards based on historical BHA figures for the same race. Indeed, they fall outside a remarkably narrow range of figures of 167.9 to 169.4 judged on the previous five years.
If Al Boum Photo had won by a wide margin, or if there had been wide margins further back, his rating from race standards would have been higher.
But he didn’t: in addition to the mere six and a quarter lengths covering the five horses named, there was just a length further back to Real Steel in sixth (despite the fact that he appeared not to stay), less than 20 lengths covering all the finishers, and Presenting Percy might have been a factor but for falling two out (you should include his estimated position at the finish in the above if you feel so).
There is one other aspect of the Cheltenham Gold Cup which is relevant to an assessment of the race and to races in general, and it is that it was run in a modest overall time for the grade.
Well-run races, completed in good overall times, reflect the abilities of the protagonists more than do falsely-run (too fast, or, more often, too slow) ones. It is good handicapping practice to take a positive view of races backed up by good times and to take a less positive one of those that are not.
Finalised timefigures are likely to be available only after a card has been completed, and sometimes not for a long time after that. But the principle still holds. Calculating timefigures (and, even better, sectional figures) would need to be the subject of a different series of articles.
There are a few other things to consider when assessing form through ratings. They include that you should regard solid handicap form more favourably than other form – assessments in such a context are much more likely to be correct than in others – and be wary of promoting a horse above what it has achieved in handicaps recently without good reason (154-rated Bun Doran in the Champion Chase is an example).
It is also true that it is easier to arrange horses in an order of ranking than it is to be sure what the precise differences in achievement between those horses are. This is the basis of “positional handicapping” and informs the idea that you may wish to keep one horse ahead of another that it has beaten, if not necessarily to the full degree to which that was actually the case.
Always consider how a race would look on ratings if it were to be run again under the same circumstances. Generally speaking, you want the horses that came to the fore to have that reflected in their ratings, and for those that didn’t to have that reflected in their ratings, too.
There are (many) exceptions, and you should be especially forgiving of horses who finished down the field at short prices for it is likely they simply under-performed. But the racecourse test is the one that matters most: if you rate a horse highly and it repeatedly fails to justify that on the track, the likelihood is that you were wrong to do so and need to amend.
Remember that it is more robust to rate horses on races (interpolation) than races on horses (extrapolation). You should learn from results as to whether a given horse does or does not act on a type of ground: you should never start your assessment of a race on an individual assumption.
At the end of the day, rating races can seem like a lot of effort for sometimes little payoff. Plenty of people do perfectly alright without running a handicap of their own (though most of them will be influenced by ratings, whether they know it or not). Ratings alone are unlikely to turn you into a phenomenally successful bettor overnight.
But there is one thing of which I am sure after many years of dealing with ratings myself: a formalised and numerical approach to assessment will improve your understanding of form and of a population of horses. Views become quantified and tested against events in a numerical and inescapable manner. A greater understanding is also likely to lead to greater enjoyment.
That, for me, has made the effort worth it.