machine learning to predict winner based on field

My only rule: If you bid now, you start now until finished. Details: I have a mysql db and I would like a script to be setup, that can monitor for a pattern called handicap my site is [login to view URL] Its a site to be able to predict winners or losers of horse racing, based on historical data. Inside of this are the meetings of each race. Example: [login to view URL] [login to view URL] Basically I'm using a calculation called handicap. I would like machine learning to be able to predict the success rate for set handicap. Then i will change the handicap figure and run the script again. --> This is one particular horse race [login to view URL] the winner is : muse (look down the bottom for results) If i sort the horses by handicap fastest to slowest (with search field (1200) ** only 4 horses ran, so we ignore (if have less than 50% of the horses in the field we ignore) 10 horses ran, so we need at least 5 entries. If i sort the horses by handicap fastest to slowest (with search field (1300) Myprayer the fastest handicap time muse is 2nd last Thor's Thunder is last If i sort the horses by handicap fastest to slowest (with search field (1400) One Of The Kind has the fastest handicap time muse is 3rd fastest If i sort the horses by handicap fastest to slowest (with search field (1500) ** no horse ran at this race, so we ignore (if have less than 50% of the horses in the field we ignore) If i sort the horses by handicap fastest to slowest (with search field (1600) ** only 3 horses ran, so we ignore (if have less than 50% of the horses in the field we ignore) What did we get: ############## For race 3 of an event We predicted that Muse: 3rd 1400 Muse: 5th 1300 Other distance don't take into account for Muse We need this for every horse within 200Metres of the set distance --> 1400M i.e --> Impatience,Thors Thunder