All advice is greatly appreciated and heeded. I take the point. Stats are one dimensional and a horse race is a 3D thing. The attraction with the numbers, I suppose, is time. If you are pushed for time on Friday nights/Sat mornings you need some reliable shortcuts. But I realise any shortcut can be no substitution for proper analysis. And Mr Turfaccountant, whose methods I was looking at, makes claims for them based, he says, on large-sample database trials (albeit several years ago now). I think the idea would be to combine various methods as so many filters. The triok would be finding a combination of factors that didn't filter out too many races.

Thanks again

Hermes