Rene Nunez

When I built v1, I wrapped the whole thing around one output: expected runs per team. The reasoning was that expected runs gives you all three markets at once, which is clean if it works. If team A projects to 6.5 and team B to 3.5, a distribution over those gives you win probabilities for moneyline, the spread of the distributions gives you the runline (which is really just a moneyline derivative), and the sum gives you totals. 6.5 + 3.5 = 10, total line is 8, that's two runs of edge, flag the over. One number, three markets, done.

When I started this project, I was operating on a personal hypothesis I'd had for a couple of years while modeling sports: since baseball has the highest variance of any major sport, introducing overly complex modeling techniques would likely lead to overfitting and, ultimately, worse performance.