HarryH
5th September 2001, 01:28.36 AM
I must admit to being a "data-freak"! I feel like I'm in "Data Heaven". Here we have a great set of software tools and more data available everyday than I ever thought I could afford. I am usually pretty good a picking out patterns out of a lot of "noise". (30 yrs aerospace experience doing this sort of thing). I have written many horse programs which scan databases of the output of a handicapping tool looking for the optimum combination of parameters ( Aplus, Multi-Strats, Neurax and others) . Always looking for that "Holy Grail".
I have found some spot plays that have worked very well but occurred so infrequently that I could follow the data every race day for 2-3 weeks with no plays, and then the one day forgot to check for that spot play, it hits (one paid $238 to win!!).
HTR and its associated database is continuing the tradition. I used to think 1000-2000 races was plenty to secure the performance of a given criteria. I have about 3 mos. of data on almost every track (I throw out tracks I have no possibility to wager on). I have found several plays that worked early on but fall through the floor as more data is added to the database. Some of the "Holy Grails" I have found here were the result of my programming errors (for example, not realising that nFIN only shows 1-2-3 finishes and testing on nFIN<2 for the winner). I have found a few "True" performers, but I fear putting them to the test with real $$$, worrying that they will "return to the mean"===> become a losing ROI overall.
I guess I have gone on long enough; I will close with an example play for others to play with and tweak; maybe it could become the "Real Thing":
Using Impact, Play Exacta boxes keying the top selection for the nESP sort and the nTRN sort (keyboard 1 and 5) : 1-2,3,4,5 and 5-2,3,4 for a max of 7x$2x2 =$28. (the 1,2,3,4,5 mean keyboard sort keys, top selection). Do not bet "Low odds" boxes. I will leave the definition of "low odds" to the user. Many times you will have multiple hits. Other filters would probably be called for (I haven't given up on it yet). I got real excited when I saw BEL rack up almost $2000 in profits before the last month's performance sank it into negative ROI territory. I did find several tracks with a pos. ROI, but the year is still young! It seems to be pretty track dependent and very "streaky". If you have read this far I applaud your patience.
I have found some spot plays that have worked very well but occurred so infrequently that I could follow the data every race day for 2-3 weeks with no plays, and then the one day forgot to check for that spot play, it hits (one paid $238 to win!!).
HTR and its associated database is continuing the tradition. I used to think 1000-2000 races was plenty to secure the performance of a given criteria. I have about 3 mos. of data on almost every track (I throw out tracks I have no possibility to wager on). I have found several plays that worked early on but fall through the floor as more data is added to the database. Some of the "Holy Grails" I have found here were the result of my programming errors (for example, not realising that nFIN only shows 1-2-3 finishes and testing on nFIN<2 for the winner). I have found a few "True" performers, but I fear putting them to the test with real $$$, worrying that they will "return to the mean"===> become a losing ROI overall.
I guess I have gone on long enough; I will close with an example play for others to play with and tweak; maybe it could become the "Real Thing":
Using Impact, Play Exacta boxes keying the top selection for the nESP sort and the nTRN sort (keyboard 1 and 5) : 1-2,3,4,5 and 5-2,3,4 for a max of 7x$2x2 =$28. (the 1,2,3,4,5 mean keyboard sort keys, top selection). Do not bet "Low odds" boxes. I will leave the definition of "low odds" to the user. Many times you will have multiple hits. Other filters would probably be called for (I haven't given up on it yet). I got real excited when I saw BEL rack up almost $2000 in profits before the last month's performance sank it into negative ROI territory. I did find several tracks with a pos. ROI, but the year is still young! It seems to be pretty track dependent and very "streaky". If you have read this far I applaud your patience.