Statistical Analysis of large data sample

[YorkshireFootyFan]

I've got a ratings technique, I've rated over 35000 games, although for the analysis i'm about to embark on I intend to use just 20000 of them which are for the Football League ie, The Championship, League 1 and League 2. The other 15000 cover the top divisions in England, Germany and Norway, but I don't think these divisions should be used for the modelling, I do think that the three leagues I'm using are broadly similar. So with a sample of 20000 ratings and the results for these games, what statistical approach should I use? I did plot them on a scatter graph and use the best fit line to produce prices for the 1x2 market - this was quite successful and if the recommendations were followed blindly, back testing suggests it would make anything from a small loss to moderate/reasonable profits each season. I have considered using Monte Carlo analysis to produce a 1x2 price for each rating increment, i'm not quite ready to do this, I will be in a day or two, I did mess about with it a little the other week though and again it seems promising. Does anyone think there are better methods I should consider using?