Moeee - Your point/concern on the size of the overlay is a good one. In short, I have a sample of about 3200 bets with the perceived overlay (this needs more explanation but will suffice for now). I break these overlays into groups for analytical purposes; 0-5%, 5-10%, 10-15% etc etc. The 0-5% group has returned a long term profit so I don't mind the 'skinny' value. I appreciate my sample is rather small but there doesn't appear to be too much variation (I have more than 40000 matches in my database but only have odds for about 3200). Funnily enough the lowest return comes at the higher 40% plus overlays (I could speculate why but there's not enough space).

Thanks for the question.