Pengo, I haven't the stats anymore, but I did do extensive research a few years ago, always frustratingly when the S/R increased due to filters the SP decreased in proportion, so same story. loss of about 10% on turnover. Amazingingly the closest to break even was backing odds on chances only.
Another thing was that there was virtually no difference (over time) in S/R no matter what the class of race, and again where there was; the SP decreased in proportion. Also virtually no difference in LOT (over time) between the actual fav or the pre-post fav.