Luck evens out over 162 games. You lose some games you shouldn't have...you win some you shouldn't have. Where you finish is where you finish. Baseball is too streaky to me to suggest that looking at run differential tells you what a team is going to do over a sample size of 30 games.
I'd be interested to see if there's any study that takes, for example, records and run-differential after 100 or 120 games, and see which is more predictive of season-end records. Run differential won't tell you what a team is going to do, but I'd guess probably tells you what they are most likely to do more than their record does.
Generally a team is conisdered lucky or unlucky depending on their performance in 1 run games. For example, if a team is really upside down (let's say 5 - 14) over the first half of the year, you should expect an improved over the second half of the year (and vice-versa).
I used to love to read baseball statistical analysis stuff(written by stat geeks). And they make a very convincing argument about using run differential, or "expected win-loss record", as a predictor of future results. Unfortunately, I don't get into this stuff anymore. So I can't really get into the details. Well, in another couple of month, we'll probably see who's right.
I'll still be right...because they'll still be whatever they are after 162 games. And that's what they'll be.
