Yeah - as I mentioned when I first joined, I created a basketball sim that was working off of adjusted off / def efficiencies (adjusted using excel's solver) and running off of basketballvalue.com's data set. I would run 10 trials of 1,000 samples based on each team's adj off / def efficiency, avg possessions and a weighted standard deviation to create a decent probability function of what might happen, and then compare the outcome of my samples to the spread.
Unfortunately, I didn't really understand how gambling markets worked. I assumed that "vegas" was skewing the numbers to collect on an ignorant public (not a bad theory when you see people citing trends and other information that gambling sites are giving to you FOR FREE as a method for making a bet). However, I came to the realization that any edge my model was telling me was probably already baked into the market (I wasn't betting openers) and that the issue was with my model, not the line. It went on a nice run to get my hopes up, fizzled for a bit, and then I quickly quit.
I was also keeping a simple "record" of wins and losses, ignoring any sort of line prediction value or purchase price, mistake number two.
I could go back with what (I think) I know today and try again, but it was kind of a pain with the daily upkeep.
Also, if anyone wants the "software" (again, all done in excel), I'd be happy to release it.