One can only get so far by looking at one race, but as expected the 3.15 Doncaster today proved interesting, with none of the top five as rated by VDW's basic ability rating making the frame and only one of the three other horses who had won races of at least the same class as today's doing so. In the event the race was won by the horse ranked 15th of 17 on ability, with the placed horses ranked 9th, 7= and 11th respectively.
VDW's ability rating has three virtues: it is undoubtedly a discriminating measure of ability (based on the simple hypothesis that a horse which on average wins races of class 100 is better than one which on average wins races of class 40); it is easy to calculate, and it can be applied as easily to handicaps and non handicaps. But for the Flat especially VDW was clear that it had its limitations, namely it does not enable one to "get a full measure" of the ability of younger horses.
In my view, for assessing Flat handicaps VDW's ability rating has three limitations:
1) as VDW himself stated, it does not always enable one to "get a full measure" of ability, not just with very young horses but also in my experience with horses as old as 5. For a common pattern is that decent handicappers improve not just from 2 to 3, but also from 3 to 4, and from 4 to 5, and indeed a not insignificant number improve furher from 5 to 6 and even from 6 to 7. This is partly a matter of date of birth, partly rate of maturation, and partly a question of experience. The winner today was the second most exposed horse in the race with 67 runs prior to today and only Advanced with 81 having had more. By contrast Head Of Steam, as a 5yo only a year younger than Thunderball, had had only 12. Plainly much more likely that the VDW ability rating will have the "full measure" of the likes of Advanced and Thunderball than of the likes of Head Of Steam and younger horses still such as Prodigality and Khubala;
2) the ability rating is career-encompassing and thus assumes, even for horses for which, arguably, it does give "a full measure", that the ability level doesn't drop. And indeed VDW does assume that: "Usually ability will remain static, fluctuate slightly, progress steadily upwards or rise sharply. Ability cannot eventually fade ..." That, of course, is nonsense. Both a horse's actual ability, and his ability rating, can and do fade: not infrequently we see horses able to contest and win decent class handicaps as 4 or 5yos running in claimers and sellers as 9yos and older, no longer able to contest decent handicaps and with every claimer they win seeing their ability ratings drop;
3) the ability rating does not lend itself to assessing the chance of a horse in a race being analysed. One can, of course, make the kind of comparison I made in my earlier post today - ie that seven out of the 17 runners had previously won races of at least the same class as today's, and the even stronger point that four of them won races of at least today's class on average when they won. But in addition to not taking account of good but not winning performances, the basic ability rating does not permit the dynamic factor of relative weight to be brought into the equation. For example, top weighted Secret Witness had won a higher class race than today's earlier this season - a class 129 on 16/05/12. So in theory today's class 104 was well within his competence. But he won that race, by 0.5l, off an Official Rating of 94. Today he was running off 105, 11lb higher in the relative weights. The Form Book records that 16/05/12 win as achieved "driven out", which hardly suggests he would have won the race had he been carrying an extra 11lb. Yet, apart from his far from unimportant weight mechanism illustrated with the likes of Stray Shot and Zamandra, VDW had no means of bringing that element into his calculations.
Not, of course, that it was easy to do so when VDW was writing his letters and articles, because only those with the indefatigability of Marvex addressed the issue, and by his own admission Marvex devoted "between 70 and 80 hours a week to the job" and "in addition I have a certain amount of assistance on the clerical side, say a total of 150 hours in all". With his full time jobs as first a radio mechanic and later in the Market Harborough hosiery trade, VDW did not have that amount of time available.
But we are more fortunate, for we have information about relative weights readily available, and that is why I reached the conclusion some while ago that using Official Ratings offered a better means of assessing the class of races, and the ability of horses, than VDW's ability rating, a means that enables each of the three limitations with the VDW ability rating to be addressed.
The method was first suggested to me by Guy Ward ("the Mathematician") and almost simultaneously demonstrated in action with considerable success by "John In Brasil" from the Gummy and other forums. The way I use it is simple: I merely sum the Official Ratings of the runners in a race, make some adjustment when there are horses running from outside the handicap, and divide by the number of runners. Thus today's 3.15 Doncaster is, for my purposes, a race of class 91.8.
One could use figures of that kind in a similar way to that in which VDW used win prize money: one could sum the average OR of each race a horse has won, divide by the number of wins, and use that for ranking the field and comparing to the class of the race being analysed. But that wouldn't deal with any of the three limitations referred to above.
But if one adds the OR the winner ran off to the class of the race won as measured by the average OR, one gets a different rating: in the case of the winner today the rating for the race would be 177.8. And one can adjust that for the winning margin, though obviously there is not much to be added for the neck by which Thunderball beat Ancient Cross.
With ratings of this kind one is in a position to take account of relative weight. To revert to Secret Witness's win on 16/05/12, that race was very similar to today's in terms of average ORs - 91.6. He won by 0.5l off 94, which gives a rating of around 186.0. We can then compare that rating to the task facing Secret Witness today - winning a class 91.8 off 105. That is, in the terms of this method of rating, to win today Secret Witness would have had to return a rating of at least 196.8 (91.8 + 105), ie a rating 10 above his last winning one.
Compare Secret Witness to the winner today. Thunderball's last winning run was on 11/05/12, a race of average OR 86.7, well below the class of today's race or that of Secret Witness's win on 16/05/12. No chance that Thunderball would beat Secret Witness, one would conclude. But the picture changes when one factors in relative weight. Because today Secret Witness was running in a race of almost the same class as on 16/05/12 off 11lb more in the relative weights. Thunderball was running in a race 5.1 higher in average OR but off 4lb less in the relative weights. Whereas to win Secret Witness had to record a rating about 10 higher than on 16/05/12, bearing in mind that Thunderball had won by a comfortable 1.25l on 11/05/12 he only had to return a similar rating. Or, put another way, one could argue that today Thunderball had an 10lb pull over Secret Witness based on their last winning runs, ie about 2.5l over 6f.
Obviously comparison based on last winning runs some five months ago offers no definitive answer to the question which of the two would beat the other - other highly relevant factors would be evidence of improved ability since May, an appraisal of the two horses' recent form, and the question of conditions (distance, going, course type, draw etc). But the approach outlined does offer an alternative means of assessing race class and horse ability which, when accompanied by a measure of realistic potential, addresses all three of the limitations of VDW's ability rating.
Least anyone should think I regarded Thunderball as the most likely winner today, not so. But in the field of 17, only Colonel Mak had put in a winning handicap run in 2012 which, when assessed in the way just demonstrated, made him a more likely winner than Thunderball.
