Return to the Main FAQ's PageFAQ # 35
ALL-WaysTM Support
How many races do I need in the Race Database for a track before I can start generating custom Handicapping Profiles?
There is no magic threshold as to the number of races required for creating good Handicapping Profiles. To be sure, the more races you have in the Race Database for a track, within reason, the better the performance that can be expected of Handicapping Profiles. But, the reason may be a bit different than you expect. Also, what do we mean by "within reason"?
We have found that Multiple Regression Analysis generated profiles can be very effective with as few races as 20 to 30 of a type. Indeed, even 10 races of a type can produce MRA profiles that do a commendable job of isolating the contenders in a race. What many ALL-Ways handicappers do when handicapping a track for the first time is to download the previous five days of racing that BRIS generally makes available and then to download the races each day for the next week. This way, they get to 90 to 100 races in the span of about a week and they can start creating Handicapping Profiles right away.
The main reason that Handicapping Profiles get better with the more races in the Race Database for a track is because you can segment the races into even more pinpointed categories. This is a very important point. Let’s say that you have a profile for non-maiden dirt sprints based on 20 non-maiden dirt sprint races. When your Race Database grows to a level where you have 40 non maiden dirt sprints, instead of staying with the 40 races, we suggest you segment the non maiden dirt sprints even further so you have two new categories of non-maiden dirt sprints, perhaps with 20 races for each category. For example, you may separate the 40 non maiden dirt sprints into higher class races (BRIS Race Rating at 114 and above) and lower class races (BRIS Race Rating at 113 and below). When you have a few more races, we strongly encourage you to segment races in accordance with the Race Pace Shape categories that we published in our October 1996 Newsletter. You can view and/or print this newsletter in the Newsletter section of our Web site. To repeat our main point here, as you build your Race Database for a track, we believe you will do best by continuing to go deeper in the segmentation of races as opposed to just having more races in total.
Now, for what we mean by "within reason" when we say, "the more races the better". This one can probably be hotly debated. We believe a Race Database can become too large if it means that the races go too far back in time. This is an unscientific and largely untested position. We do suggest that you be wary when your Race Database for a track gets much above 1000 races.