The totals market in soccer

With hockey dead for the next few days, I decided to maybe look into my resolution to no longer pick soccer games/outcomes and determine if there is not a strategy I could use in picking over/under in soccer games.

For assistance, I've first looked into a journal article called "ARE BEHAVIORAL BIASES CONSISTENT ACROSS THE ATLANTIC?: THE OVER/UNDER MARKET FOR EUROPEAN SOCCER" by Rodney Paul, et al.

Coming from the Journal of Gambling Business and Economics (September 1, 2009), Paul looks at several European Leagues from 2005-2007 to determine if the under is underbet in soccer. He does determine that it is and that, as a strategy, if a bettor were to bet theunder in a soccer game, they would lose less money overall than if they bet the over. Not the greatest endorsement to attempt to find a strategy in picking soccer totals, but if the under loses less often, then perhaps we can find a way to target the unders most likely to occur.

So, using data from football-data.co.uk , I've determine how well the under performs as opposed to the over. Well, for this year, I don't know if somethign special has happened since 2007, but the over is doing MUCH better than unders. Overs, generally, are occuring as frequently as unders, but winning(losing) more(less) units. Only a handful of leagues have Unders markets that are fine. Generally speaking, it appears the under is actually valued at lower odds than the overs. So even int he leagues where the under occurs pretty frequently, unit wise it does pretty terrible.

Here's some data, so far, of the more interesting numbers:
EPL: Over/Under is 52/48%, but units are -6.5/-21.5
English Championship: Over/Under is 52/48% but units are +0.96/-41..2 (wtf, right?)
English 1: Over/Under is 54/46% but units are +6.73/-42.49 (crazy)
English 2: Over/Under is 49/51% but units are -9.63/-28.7 (whargarbl!)
Scot 2: Over/Under is 62%/38% but units are +6.7/-17.96
Bundesligue 1: 63/37 and +11.38/-33.7
Segunda Division: 45/55 and -9.31/-22.98
Serie B: 43/57 and -17.38/-17.21
French League 1: 42/58 and -11.59/-14.18
Portugal: 47/53 and +.81/-17.28

God help you if you were following Paul's observation this year. Judging by how low the odds are of an under happening, it looks like bookmakers are finding people now betting on the under more than the over. So, if that's not the way to go, perhaps I can find a way to mine out the likely over/under. I'm thinking of looking into methods involving poisson distribution of goal differential between 2 teams.