I think if you flip a coin 100 times, you're going to end up with 50 heads/50 tails on the dot, relatively few times. But I bet you end up with between 45 and 55 heads, about 90% of the time (or even more), and I think that's all the stats/science is trying to do, is give you some predictability within a reasonable variance. I think that (in general) the stats do a relatively good job of giving us predictive information, and they also do a good job of telling us "this is what happened". The problem is when people say "hey, this team 8 years ago scored less runs than they allowed, and they finished X games over .500". Obviously outliers occur. It doesn't mean that the predictiveness of the math is wrong, it just means that's one of the outliers. It doesn't hurt to look at the outliers and look at the information we have and see if there's some causation for why the statistical information doesn't add up to the W's and L's for that particular season. Is it just luck? Is it a combination of good bullpen use and luck (good record in 1 run games?) 162 games, by some measures, is still a small sample size. The Brewers this year have league average OPS and league average team ERA. yet they're 8 games below 81 wins. Why is that? Bad bullpen management? Bad luck? Did they create too many outs on the bases, depriving themselves of runscoring opportunities? It's worth looking at, not to say, "the statistical information is wrong", but to see what caused a minor outlier in the data.