Things That Don't Work
The trouble with researching handicapping ideas is that most things you look at don�t work. You spend hours studying some arcane factor and you come up with zilch, a fizzled factor that has little or no impact. Why share these with you? Well, if you know what not to do, you won�t waste your time and money trying it. Also, each research dud tends to generate ideas for new potential spot plays or significant predictors.
Things That Don�t Work
Already Won Today Jockeys: I heard a trainer/jockey handicapper�s speech where he mentioned that if you played all races of jockeys who had won already that day, it was a break-even proposition. The idea was that these jocks were probably riding well and often at the start of a streak. It sounded credible to me, so I tested it. Here are my results:
Of 763 performances where the jockey hadn�t already won today, there was an 11.5% win rate with a .66 ROI. Of 233 performances where the jockey had already won at least once on the card, there was a 13.7% win rate with a .72 ROI.. Both of these subsets had below-average ROIs. (I hear the question now: Since these subsets combine to form all races within the test period, how can they both have below-average ROIs? Because, I answer, tracks can go through periods of low payoffs, which will generate below-average ROIs.) Even if I reran this study in a period of higher payoffs, the already-won-today jockeys don�t look like a good prospect for an edge-generating predictor. So, betting on a horse just because his jockey won already today is a useless endeavor.
Now, I suspect that jockey streaks are a useful predictor and I just haven�t found the right way to measure them yet. Maybe the jock�s A/E Ratio over the last 30 mounts... hmm.
Trainer Meet Win Percentage: I wondered, at one point, whether a trainer�s win percentage over the current meet was a useful predictor. The Racing Form is kind enough to include that stat right in the past performances � it would be rude of us not to use it. So I did a study of different ranges of trainer-meet win percentages: 0%-4%, 5%-9%, 10%-14%, 15%-19%, etc.
There appears to be little correlation between a trainer�s before-the-fact meet win percentage and the percentage he wins afterwards. For instance, trainers with a 30%-34% meet win percentage won 18% of the time in my study, while trainers with a 10%-14% rate won 16% of the time. There seemed to be a tendency for these trainers� performances to regress to the mean � in other words, those who over-performed tended to do less well, and those who under-performed tended to do better. This didn�t happen uniformly for all the groups, though with a larger sample, it might have.
These fizzled studies often point me towards further research that might be useful. One possibly significant finding was that the trainers with a meet win percentage of 0%-4% did continue to under-perform with a 7% win rate. (Unfortunately, I didn�t track average odds, so I can�t do the A/E Ratio calculation which would tell me if this low percentage was simply because these horses were longshots.) It might be possible to eliminate or downgrade low meet-win-percentage trainers.
Things That Might Work
The Prancing Horse: When I was physically going to the track every day, I did a short test of my aptitude at recognizing the body language of horses. I had several preset descriptions which I would mark down as I watched the horses in the paddock, such as excellent, good, okay, dull, prancing, muscled, dappled, flat tail, etc. I then tracked how the horses I designated in these ways did. During that time, I marked 19 horses as "prancing." Of those 19, 8 won, for a 42% win rate and a 2.27 ROI. Tiny sample, I know, but encouraging. It included a $26.60 horse and an $18.00 horse.
Body language is a subjective thing, and that�s part of what gives it its potential edge. If you have a knack for it, it�s possible to gain an advantage because you�re one of the few people seeing it. Also, it�s so time-sensitive � it�s something that�s noticed a few minutes before post-time. There�s no time to disseminate it to the crowd. Maybe I ought to get that old Joe Takach tape out again.
Handicapping the Handicappers: I used to take a certain Southern California handicapper�s picks in the local newspaper and track them using wagering analysis software. I soon noticed that this guy was exceptional at high-class races such as stakes or claiming races of $40,000 and up. I began to bet his picks in these races. In the time that I did it, he hit 17 of 55 races, for a 31% win rate and a 1.08 ROI.
One character flaw I�ve had to battle in myself and have seen in other handicappers is the tendency to find something that works, and then abandon it for no special reason. Some handicappers like the search for profitable methods more than the drudgery of trying to make a buck with them. I�ve pretty much cured myself of this, but these are a couple of handicapping methods I may need to revisit.
This is the behind-the-scenes stuff that every handicapping researcher has to deal with. For every nugget you find, there are several lumps of coal. Uh-oh, it's post-time � time to leave the things that don�t work and the things that might work, and with a little luck, get back to the things that do work.NC
Copyright �2001 NetCapper Inc. All rights reserved.