HH, I think a statistician will tell you that 'regression' works in both directions, and thus, there's no need to bother with having to say 'progression' to the mean. In other words, 'regression' simply means to move toward the mean; you can 'regress' toward a better number or 'regress' toward a worse number. Of course, that's different than the 'every-day' definition of regression. lumberrule, Ennder is correct when he says 162 game samples have accuracy issues. You really need about three full seasons for statistics to be able to stand on their own. That's one reason why splits can break pretty badly; it can take much more than three seasons to get a reasonably useful sample to tell how a guy's going to perform vs. lefties, for instance. That doesn't exclude bringing in other information to help get some meaning out of smaller samples. As an example, Prince struck out an awful lot of times in his first two games of the 2005 season. Statistically, that's meaningless. However, the numerous observations from people saying, "he didn't have that lousy approach in spring training," filled in some important blanks.