Trainer Stats (Part 2)
Trainer stat handicapping seems like it should be so easy. You go through, highlight all the positive stats for each trainer, and head for the windows. The modern profusion of data and databases has led to a new type of handicapping sometimes referred to as database handicapping, of which trainer stat handicapping is a part. This, of course, has led to a whole slew of new ways to lose. Trainer stat handicapping isn�t simple � here are some of the reasons why.
What�s he done lately? For any statistic, it�s important to look at both the long term and short term results. For instance, as of July 11, N. Hines had an overall record of 19/116, 16%, 0.88 ROI. Not bad, a bit above average. However, Hines� record at Hollywood Park was 9/39, 23%, 1.44 ROI. Hines was in a hot streak. I�m a big believer in hot streaks for trainers, jockeys, horses and even handicappers, for that matter. Our histories don�t comprise a discrete series of mathematically independent events, as some handicapping experts would assert. Sometimes we�re humming along on all six cylinders, other times we�re in the shop with our transmission all over the garage floor. It�s useful to know what a trainer has done lately. What he did five years ago may not be all that relevant anymore.
Short term stats can be kept in different ways: the last 10 starts, the last two weeks, the current meet. It�s meaningful information: is the trainer in a hot or cold streak? With the caveat that cold streaks are less significant (see last week�s article), it�s like the old Wall Street adage, "Don�t fight the trend."
What were the average odds of the horses in this sample? Say a trainer has a record of 2/40, 5% wins. It seems like a pretty negative stat, but is it? The ROI gives us a hint. If it�s 0.37, that�s not so good; if it�s 1.20, that�s better. But in fact, even with the ROI information, we�re still pretty much in the dark. A fluke longshot winner can turn an ROI positive, and it really doesn�t have much predictive value. You need to know what the average odds of the sample were.
For instance, if a trainer�s win% is 5% and his average odds are 20/1, the win rate is explained by the odds. If this trainer has a 5/2 horse going off in a race, it may be a good play. However, if this 5% trainer�s average odds are 6/1, that�s a clear sign that he is under-performing. It comes back to one of my favorite subjects: the A/E Ratio. Unfortunately, I don�t know of any stat service that provides this information. It�s an important thing to know, but you�ll never know it unless you keep your own stats.
An after-the-case statistic is different than a before-the-case predictor. So many handicappers don�t understand this. When you see a nice juicy positive trainer stat � here�s one: Jack Carava is 4/9, 44%, 2.81 ROI if his horses are wearing front wraps � you tend to think that means that Carava�s front-wrapped horses in the future will win around 44% of the time for a big profit. No, that�s not necessarily true at all. Just because a subcategory was positive in the past doesn�t at all mean it will continue positive in the future.
It�s vital to have not only statistics, but statistics on your statistics. Stay with me here. In other words, you mark down your positive stats before the fact, and then keep track of how you would have done if you had bet them. That way, you are treating them as before-the-case predictors, not just after-the-case statistics. You�ll be surprised how many after-the-case positive statistics do zilch as predictors. But not all. When you find a positive before-the-case predictor, then you�ve got something.NC
Copyright �2001 NetCapper Inc. All rights reserved.