Quote:
Originally Posted by justin13892002
To answer FTL's question first. I have right now 181 races in the database, they are over 5 tracks, Aqueduct, Woodbine, Gulfstream, Keeneland, and Hollywood. I do have each track separated to create a model for each specific track.
I have 4 designated COLUMNS, Top 3 SP Screen, Top 3 EXDC Screen, Top 3 Thoromation, Top 3 Total Energy.
What I did was enter the results of each race, and the Numbers "Change colors" To let me know where the W-P-S horses finished.
What I have found is some races certain factors seem to do better, as I said 1 race Top SP will win, yet wont be ranked anywhere else. Sometimes the horse ranked in All 4 will win. Its very tough to distinguish which factor will be the most important, not so much finding the "Win" Horse, but just finding the Top 2 contenders or so.
I appreciate your guys help... :-)
|
Ok, so on average, you have about 36 races per track. If I understand correctly, you have all distances, surfaces, sexes and race types combined into one model. If this is true, you have two problems. One, you need to break these races down by at least the classifications I just mentioned and two, you need more races (data) for each track.
In my experience, maidens, claimers and allowance horses will run differently from each other. Also, males and females will run differently as well. And all of these can, and usually do, run differently at various distances.
It could be for the above reasons that your current models appear "jumbled". My suggestion would be to break you models down by distance, surface, sex and race types. Then add more races to your database and see if each model falls in line, thus giving you a more accurate picture for each model.
As a side note to you and anyone else reading this, it is for the reasons I suggested that I confine my play to tracks with long meets. Tracks such as Del Mar, Saratoga, Keenland, Churchill Downs and numerous others with short meets make it difficult to "get a line" on how the horses are running. By the time your models are starting to become meaningful, the meet is over.
As an additional side note, at tracks that card races for a whole variety of conditions, such as NW2L, NW3L, NW4L, NW16m, NW26M, NW36M, NW1Y, NW2Y and etc., I keep separate models for each of these conditions in addition to those classifications noted above. It requires a lot of data, so a long meet is very helpful. You would be surprised at how these models can differ. The best model I ever had, was one where the #1 or #2 first fraction horse won, as I recall, 34 straight races, many at very good prices. Keep in mind, that at the same time, other models were NOT running the same way. It's a big edge when you know what you are looking for.
Good luck.