Zoe,
I don't see how you can think a larger database doesn't help in formulating systems. Taken to extremes if you have a database of one race you might find that barrier 5 at track x wins 100% of races, an obviously false conclusion. With a database of 100,000 races you would probably end up with a reasonable sort of curve biased towards the better barriers. Something you may want to use as part of a system.

Dale and Bagwhan,
I tend to agree with Dale's view of the word retrofitting and that was what I alluded to with the "letters in a horse's name" system. I'm sure you could retrofit a reasonable number of races backfitting an assortment of letters in an assortment of different positions until you get some impressive results. But having said that I suppose we all retrofit in a way to try to invent systems, even handicapping is retrofitting of a sort, so Bagwahn's view holds water too.

Chrome and Marcus,
Thanks for those posts. Interesting and informative.

KV