Just ran across this and thought it was pretty interesting. Based on 10,000 computer simulations, Valpo wins the league 48% of the time. Fun read for the guys who are into that sort of stuff.
http://nycbuckets.com/2013/01/10000-sims-horizon-league-is-wide-open/ (http://nycbuckets.com/2013/01/10000-sims-horizon-league-is-wide-open/)
But only 31% outright. An additional 17% of the time Valpo got a share of the title. And 9% of the time their was a tie for first _not_ involving Valpo. Winning 0 times in 10,000 simulations? Milwaukee. Going winless in 9% of simulations. Valpo goes undefeated 13 times of 10,000. Valpo with 10.8 "projected" wins to Detroit's 10.4, and YSU's 9.1.
Good fun. But funny that he doesn't say what he's using to seed the simulations. Looks like Pomeroy's Pythagorean more than anything else I checked. But, of course, not every ranking has Detroit (or even Valpo!) that high.
Why would he share what he used? Then you could run it yourself and wouldn't have to read his site.
Well, even if you have the full details, coding the thing's not so easy.
Or, he could at least give us a few of the details. Telling us what he used for initial rankings doesn't say anything about the probabilistic engine he used to establish the result of each game.