Return to the Main FAQ's PageFAQ # 39
ALL-WaysTM Support
I have seen a case where a Multiple Regression Analysis generated Handicapping Profile does not yield great results when used in the Database Run Analysis Report. How can that be?
One of two things can cause the win percentage and the ROI to be less than expected when running a Database Run Analysis Report.
First, more often than not, this is caused by a mismatch between the races being analyzed and the races that the Handicapping Profile(s) were intended for. For example, you should not use the Multiple Regression Analysis Default non-maiden dirt sprint profile in a Database Run Analysis that includes both maiden and non maiden races. If the profile is aimed at non-maiden races, then the Database Run Analysis must be run for only non-maiden races. Another example; If the Handicapping Profile is aimed only at maiden dirt sprints where the winner paid more than $15, then the race screens must be set up so the Database Run Analysis only analyzes maiden dirt sprints where the winner paid more than $15. This mismatch between targeted and analyzed races is almost always the cause of less than expected DB Run Analysis results.
There is a second cause of the results being less than expected and that is that the results are real. Indeed, you will come across certain types of races run at certain tracks where selecting winners is just plain tough. For example (hypothetical), non-maiden dirt sprints may be tough to handicap at River Downs. The way to try to solve this problem is to segment the races into more specific sets of races. In our example, you may want to have different profiles for 6 furlong races and 6 ½ furlong races. Or you may want to divide them into higher and lower class (Race Rating) groups.