Follow BallHype on Twitter
The Hardball Times
found this 11/12/2008
-
MGL David, in your regression analysis, when you looked at the effect of experience, you held the pythag w/l records constant? i.e., you did a multiple regression? I ask this because we would expect that the players with more playoff experience would be better players.
Even if yes, I'm still not buying your conclusion until I see "why?" At the very least, I want to see how players with lots of experience did in the post-season relative to their regular season as opposed to players with little experience. The assumption (if you buy David's conclusion) is that players with post-season experience will "outplay" their lesser experinced counterparts during he post-season. If they don't, then I would want to see the "third-order" win expectancy of teams with and without post-season experience as compared to their actual winning percentage.
The reason I want to see that is that if players with more experience don't actually perform better in the post-season, then the alternative explanation if we accept that their teams win more often, is that they are providing some intangible other than actual improved performance, that enables their teams to win more often than the team's underlying performance would suggest.
I have always wanted to do a study looking at this "conventional wisdom" but I would have started, as I state above, with performance during the post season (as compared to performance during the regular season) regressed against prior post-season experience, and then gone from there.
For example, if it could be shown that players with more experience simply perform better than expected in the post-season, then our inquiry is pretty much over. If not, then we'd have to dig deeper on a team level, which has a lot more noise in it. David, you started with something very noisy (team-level win loss records, or series records, which is even more noisy).
-
David Gassko Yes, it was a multiple regression. I don't necessarily disagree with what you say, but the problem with looking at player-level stats is that you then have to adjust for league and aging effects, which can be tricky and its own level of noise. -
MGL Sure, but I don't think that the noise is going to be nearly as pervasive as when looking at the team w/l level.
Anyway, at least this should pique interest and spur some more research.
-
matts This probably is just omitted variables bias.
Teams with good bullpens, for example, are likely to succeed in the playoffs (as shown by Baseball Prospectus). Teams with good bullpens are likely to outscore their pythogrean records as well. Therefore, these teams would be more likely to make the playoffs than expected by looking at pythagorean records, and more likely to succeed in the playoffs by looking at pythagorean records. Hence, if you regress on playoff experience and pythaogrean records, excluding bullpen performance from your independent variables, the playoff experience term will capture some of this effect and falsely attribute it to the experience.
More generally, teams that play in the world series or any playoff series are not simply there by chance. If a team is in the playoffs, then ability to outscore opponents is part of the reason. Hence, pythagorean record will contribute. So will other factors, such as a strong bullpen. The lower a team's pythagorean record, the more likely these other factors are significant, and the more likely they affect both playoff experience and playoff performance, without actually implying the former causes the latter.
The best way to test an idea like this would be to test this model under many specifications with many sets of independent variables. If playoff experience continues to show up as significant, then it is likely casual.
Post-season experience matters?
THE BOOK--Playing The Percentages In Baseball —
Apparently, it does:
But while a team’s record did not have much of an effect on its odds of winning a postseason series, its advantage in experience did. A team that had, on average, one more year worth of postseason experience than its counterpart, is expected to win their post-season series 54.2 percent of the time, which is nothing to scoff at.
Wednesday Applesauce
Amazin' Avenue —
... At The Hardball Times, David Gassko looks into whether experience really does matter in the playoffs. There isn't a ton of data to work with (105 seven-game series since WW2 were considered), but the results might surprise you. And by you I mean me. ...
