|Posted: March 15, 2004, 5:58 pm - IP Logged|
I never had to get past backtesting with this... I kept detailed records, and the numbers picked were almost identical to the excel forecast function, no better, no worse... time to throw NN's on the back burner... but I must say, RAPID development is one very cool plus of the Java based JOONE engine ;-)
Data Mining... Neural network based classification has failed on several automated attempts, using both my own modified formulas (on open source excel VBA programs) and using several demo version Excel add-ins... Not totally giving up, but it's back burner for now... attempting to classify on my own with no CPU help is turning into a good thing... It may not help with accurate predictions, but I think I am almost ready to throw my hat into the trapping ring... based on classification results, using just about ALL of the statistics and other data gathered...
Playing more than one ticket per game is betting against yourself.