I think this has been the most interesting thread since I joined the Forum. Interesting, and somewhat depressing. I read the post by Syllabus23 at least a half a dozen times and sadly I could see a lot of truth in it. I say sadly because all my punting is done according to the selections my computer spits out.
I earn my living as a database programmer and have always maintained that with enough data, a fast enough computer and Don Scotts "Winning" at my elbow I should be able to get enough of an edge to beat even the TAB. What Syllabus says is true, more often than not my top selection is the TAB favorite and my ratings usually follow the final TAB divvys pretty closely. Every now and then though it picks a relative long shot. Mercedario at Belmont last Saturday for example was the highest priced success I've picked in the last several months, it paid 11-1. Also, the computer seems to pick up false favorites quite accurately.
The only place I bet at the moment is WA, simply because they are the only people I've found who are kind enough to put the last 8 years results on the net in a nice computer friendly, downloadable form (If anyone can point me to any other comparable sources of data I'd be very grateful). This (8 years) gives a computer enough data to form statistically reliable results.
I'm only a relatively small time punter but doing the unthinkable of having a level stakes bet placed hours before the jump in every race (although not every course, some I just can't crack) I come out about 8 - 10 percent ahead overall (with the assistance Maxidiv gives over the TAB) and have a strike rate of 29 - 30 percent winners.
I think you're right Syllabus, Don Scott would go mad nowadays flicking through his card index looking for a winner but I still say it's possible to pick horses purely by rating them and not lose your shirt.