|
Questions
By Ed Bain
From the February 2001 issue:
When I first started tracking trainers, I did not know what information I would find from 134 different trainer moves, but I knew I wanted percentages. As I recorded these trainer moves, “Layoffs” and “Claims” began to stand out while most other moves were low percentages or no percentages. I found myself anticipating the “Layoffs.” It was apparent that some trainers had really high percentages and some did not. So I decided to drop a 132 of those trainer moves and concentrate on two, “Layoffs and Claims.” When I did, I found I could ask myself questions and that I could then record and get answers to these questions. The first question was “How does a trainer perform returning the horse from a layoff when the horse was not sick or injured?” By recording the layoffs I now had that questioned half answered. I then asked, “Did he have a higher percentage on his sprinters or routers?” After doing the manual recording on that, I then asked, “Where is he likely to get his highest percentage from his first four races back from a layoff on either his routes or his sprints?” To answer those questions it took a massive amount of work to first organize, then set up the manual recording to find this information, along with the daily entries to keep it updated. My first way of betting with “Layoffs” was to bet on the highest percentage in each race. I quickly learned this is not a good approach. It’s like betting the fastest horse in every race. So I asked another question, “What is a good percentage?” and I decided on 30 percent. But I knew I had to include a specific number of wins with that 30 percent. I decided four wins was the best number. The reason is acquiring data was not big in the beginning. So four wins or more plus a 30 percent win rate or higher would be the platform to base my wagers. I felt four wins would reduce the luck factor and hopefully make the trainer plan out his races to produce a win. In ten races he could get two wins by design and get lucky on one for a 30% hit rate. But over twenty races he may only get two more and end up with a statistical trap. But four wins plus a 30 percent-hit rate is a 13 race sequence. At four for fourteen and 29 percent, I could then pass and avoid a lot of statistical traps with a small amount of wins and this is the reasoning I use for the 4 + 30 Play. I could also build into the trainers who needed time because of small stables but would end up to be good betting opportunities because the small stat would end up large over time. When I decided on the 4 + 30, it established a structure to my betting and I became more organized in my approach to percentages. I knew the reason was these were individual and specific trainer moves and percentages are not subject to the law of averages which says that the difference between the observed value of a sample and it’s true value will diminish as the numbers on the count increases. I felt this would be true if you’re playing every horse returning from the layoff. I also knew the individual contributes to the overall statistic and eliminates the argument about the law of averages. But it would not mean that a trainer’s hit rate would remain fixed. It could move below 30 percent or rise much higher than 30 percent with winning and losing streaks. The 4 + 30 reduced the losing streak to a manageable level. But any losing streak can be painful. But by using the 4 + 30 as a guide to make or pass bets, I could view a statistic of 4 + 29 and pass the bet. Not betting has value. Betting with the 4 + 30 has a degree of certainty and has value. And in this race, with this statistic somebody is going to win. The 4 + 30 can be measured with the bell curve. The main purpose of the bell curve is to indicate not accuracy but error. The bell curve shows the largest numbers clustered in the center, then slopes symmetrically downward with an equal number on either side of the bell. Another term would be the mean. Statisticians will tell you that 68 percent of all observations will fall within one standard deviation of the mean or the average. And 95 percent will fall within two standard deviations of the mean or the average.
All layoff trainers with a 4 + 30 stat are not alike. And neither are the horses. But all the layoffs may be similar.
The average hit rate for all layoffs is 12 percent. An individual trainer who produces a 30 percent average on his layoffs means he is not average. 12 percent is average and most trainers would fall into the average trainers’ category of 12 percent and many would be below average but would not be labeled object failures because of the bell curve, which evolves around an average. But even with a 30 percent-hit rate, there will be a 70 percent loss rate. In racing the payouts from the win mutuels make the loss rate acceptable. But 30 percent also puts this trainer in the 5 percent category when compared to the bell curves second standard deviation of the mean or average.
In August of last year I turned my stats over to HDW and they customized the “Layoff” and “Claim” statistics by arranging subsets under the primary or applicable statistic.
The subsets are 26 items the trainer is performing from the applicable statistic. These are to produce an even more specific stat. I also identified all subsets that had 4 + 30's. They can be used as support to place bets on the “Layoff” or “Claim” percentage. There are some 4 + 30's in the subsets that can be excellent reasons to place a bet. Even though the “Layoff” or “Claim” may not qualify as a 4 + 30. As an example, first time lasix, or stakes to allowance, or MSW to Maiden Claim. But for the most part I use the subsets as a reason to bet with a 4 + 30 Layoff or Claim as well as not using the subsets to place the bet. When I find the subset with a low or no-hit rate, the subset percentages are important for placing and passing bets. But I always use the primary stat as my focus.
The question that needs to be answered about wagering with percentages is “Can you win and produce a living betting on numbers that are not related to the speed of the horse?”
Speed players have a tendency to overweigh recent evidence and pay no attention to a percentage. So they lose sight of or place no importance on that percentage. After all, we know a lot more about what is happening right now than we can ever know about what will happen at some uncertain date in the future. But the evidence of percentages proves that question can be answered, yes.
|
