06/28/2002 12:00AM

Avoid claimers 'too good to be true'


LEXINGTON, Ky. - At the request of a reader who had attended one of my seminars, last week I did some research to try to discover how horses who are claimed out of a race performed in that race. I checked the results of a sample of 550 horses who had been claimed, and sorted them into different odds categories to determine whether or not there was any correlation between the betting public's estimation of their chances versus the opinions of the horsemen who did the claiming.

If there had been no correlation at all, I would have found that these horses followed the broad trend, which is that with no handicapping involved, the heaviest favorites show the smallest losses per $2 bet. The higher the odds, the larger the losses tend to be.

I was fascinated to find that the results turned out to be the opposite of what that trend suggested they should have been. The horses loved most by bettors - horses pounded down below even-money and also liked by trainers who had claimed them - did worst, with a terrible $1.31 return on investment for every $2 bet.

The ROI on claimed horses sent to the post at odds between even-money and 19-10 yielded a sub-par $1.55 ROI. The returns rose on the 2-1 through 9-2 group, which showed a $1.72 ROI. Horses at 5-1 through 9-1 did quite well, with a $1.80 ROI. And the biggest surprise was that horses at 10-1 and higher showed a small profit at $2.05.

My theory was that the horses who appealed most to bettors, and who also were attractive enough to horsemen to make them want to pay a chunk of money to purchase them, probably fell into the "too good to be true category."

The merits of such horses were so obvious that they probably should not have been running for the "bargain" claiming price they were entered for. Their connections probably knew that they were likely to be claimed away from them, and yet they didn't mind, because they had already gone off form, or were expected to do so in the near future.

Trainers were most often attracted to horses at low odds, but to the extent that they found merits in horses at higher odds, those horses outperformed expectations in the races in which they were claimed.

As promising as all of this turned out to be, it would not help bettors unless that information was made available to them. Since that might not happen any time soon, I wondered during the past week whether or not the fact that these horses appeared to be too good to be to true to bettors might suffice without the benefit of knowing if they had been claimed. Although there would be no way to capitalize on the possible upside by predicting which longshots would be claimed, and run better races than they had been expected to, there might still be horses to bet against at low odds if it turned out that bettors were just as likely as trainers to fall for the "too good to be true" phenomenon.

I used every claiming race from the same four issues of the DRF Simulcast Weekly I had used to gather the data the first time around. Since there are many more horses who are not claimed than those who are, the sample size was huge. It took a few nights of staying up well past midnight, but when my work was finished, I had accumulated a group of 2,904 horses who had run in those claiming races.

The 83 horses bet down below even-money did very poorly again, with a $1.42 ROI. The 355 runners at even-money through 19-10 also fared much worse than they had been expected to, with only a $1.47 ROI. The results improved dramatically with a fairly normal $1.72 ROI from the 1,332 horses in the 2-1 through 9-2 range. The biggest difference comes in the group of 1,134 runners at 5-1 through 9-1, who yielded a similarly unremarkable $1.69 ROI. Input from trainers might have been helpful in this range, when it is compared with the $1.80 ROI from the original sample of claimed horses, but we'll have to do without it for now.

My conclusion is that bettors are just as likely to fall into the trap of being attracted to horses who are likely to underperform at low odds as trainers are. Since that is the case, we can benefit from a larger number of horses to bet against at underlaid odds, since we have been freed from the limiting factor of the need for a claim.