I can't be more helpful either. I spent ages looking for such a database at a reasonable price. Costs plenty. Or it will cost you years in time to compile your own.

I once spent several months designing a software program that took in data and spat out selections. I added dozens of factors. It took forever to add the data and then, at the end of it all, the program usually selected....the favourite.

I wonder at the validity of testing ideas over 20,000 races anyway. My proceedure is to develop sound principles (strong handicapping ingredients) and test them on 150 recent races. Then batches of 50 races from previous years. If it holds up I test on at least 100 live races on paper. If it still holds, I start with small bets.

Its alarming the number of systems that work on recent and past races and then fail on live races.

In my own experience I find that bigger databases don't necessarily yield better results. The thing that helps is a better understanding of principles (handicapping rules) from the outset. For example, beginners at this will often look at win strike percentages and include that in their selection systems. But in fact, as experienced system developers know, place strike percentages are far more important and a more reliable guide to winning potential. No database however giagantic will turn win strike percentages into a meaningful statistic. If your theory is flawed a bigger database won't help.

Hermes