Follow BallHype on Twitter
The Hardball Times
found this 11/20/2008
-
MGL Now I'm starting to get convinved, David! Maybe.
I don't know that you are doing "pure cherrypicking." In fact, I am pretty certain that you are not. If you had used 4 years or 3 years or 6 or 7 years as your cutoff point, and you got a completely different result, then I would call that "pure cherrypicking." But, without doing the math myself, my guess is that any boundary you choose will show a difference, although maybe no so large as when you use 5 years as your boundary.
Some kind of regression of years on "regular/playoff difference" would be nice.
Are those PA numbers, the total number of reg season PA? How about the number of post-season PA for each group? I'd certainly like to get a feel for the sample sizes of each group.
I'd love for someone to speculate on why it might be the case that experienced players appear to do better in the post-season. I still find it a little hard to believe. I mean, if there is such a significant effect here, which it appears there might be, why don't we find similarly dramatic effects in other situations, like clutch? I would think that if playoff experience can magically transform an average player into a very good player and/or vice versa, surely we would see other dramatic effects in other areas, like clutch. But we don't.
I'd also like to see how age might play a role in this. Could it be that older players do better in the post-season and that it has nothing to do with post-season experience? I would think that the more post-season experience, the older the player, on the average.
I would also like to see how the regular/post-season differences match up with team w/l records? IOW, does it appear that most or all of the team w/l differences are a function of the differences in individual performance, or is there still something going on at the team level that does not show up at the individual level?
Good work.
-
David Gassko A lot of stuff, Mickey, but I'll try to cover it all.
Some kind of regression of years on "regular/playoff difference" would be nice.
What kind of regression do you suggest? What would it show beyond the results I found in this article?
Are those PA numbers, the total number of reg season PA? How about the number of post-season PA for each group? I'd certainly like to get a feel for the sample sizes of each group.
The PA numbers are post-season numbers, which is what I used to weight the difference. Technically, you're supposed to use the smaller number of the two, but I figured 99% of the time, a player has fewer PA in the playoffs than in the regular season, so I just went with that.
I mean, if there is such a significant effect here, which it appears there might be, why don't we find similarly dramatic effects in other situations, like clutch?
I would imagine that the pressure a player experiences in the postseason is much greater than the pressure of being up late in a close game. The latter happens often, the former is a rare event on a much bigger stage.
Could it be that older players do better in the post-season and that it has nothing to do with post-season experience?
Perhaps. It would be tough to disentagle age and experience, since they would obviously be highly correlated, but we look just at age and see if it shows an even stronger pattern.
I would also like to see how the regular/post-season differences match up with team w/l records? IOW, does it appear that most or all of the team w/l differences are a function of the differences in individual performance, or is there still something going on at the team level that does not show up at the individual level?
Good question, I'll try to look into it.
-
-
bsball David,
Thanks for reporting on this. You piqued my interest and I started looking at individual players to see if this makes any sense from a player perspective, i.e. do players appear to get better at the post-season as they gain experience. I started with Jeter and Williams because they have the most games played (and each makes up ~ 6% of the total experience for players with more than 4 years experience).
Jeter goes against the group results in that he gets worse after 5 years, both relative to the 1st 5 year post season and relative to regular season.
Bernie follows the group pattern (4% worse in yrs 0-4, 1.3% worse in yrs 5+). But the reason does not appear to be that he's getting better in the post season so much as that he gets a lot worse in the regular season.
Bernie Williams (OPS)
years Post Regular Diff
'95-99 .904 .945 -0.041
'00-06 .828 .841 -0.013
This is looking at OPS.
Any chance you could look into whether Bernie's situation is the norm for players in the 5+ years group? In other words does the data suggest that as players age they lose a lot more in the regular season than they do during the post season.
-
David Gassko Logically, it would have to. As they age (past a certain point, around 27-28), players start to decline, but if the differential between their numbers in the playoffs and regular season decreases, than their playoff numbers would have to decline less than their regular season numbers.
-
-
MGL "What kind of regression do you suggest? What would it show beyond the results I found in this article?"
A correlation from a linear regression, for example, always shows a little more (or at least somethng different) than an "interval comparison" which is what you show. For example, one or two players can greatly affect an "interval comparsion" and make it appear that there is a general 'correlation" when there is not. It also takes the potential "cherry picking" out of the equation.
For example,
Here is a series of data pairs:
+10 -1
+3 -5
-6 -66
+4 0
-5 1
-11 -1
There is probably little or no correlation if you ran a linear regression. Yet, if you showed an interval comparsion, you might get this:
< 0 -22
> 0 -2
making it appear as if there were a "correlation."
If you has this series of data pairs:
-11 +3
-9 -22
-3 +11
-1 -4
1 -6
3 +9
8 -3
13 -12
Again, we have little true correlation, I would think. If we used the same intervals as above, our interval comparison would be:
< 0 -12
>0 -12
Looking like, again, no "correlation."
But, if we "cherry-picked" our intervals, we could say:
< -4 -19
>-4 -5
Looking like a "correlation" exists.
That was all I was trying to say, if that makes any sense.
-
bsball David,
Could you ook at the results for only the players that contribute to the 5+ category, but looking at both periods (<5, 5+)? Do these players show improvement as they get more experience or is it a case that they represent a different population to the rest of the group (i.e. players with no experience in the 5+ group).
I am wondering if you may have something like this.
<Jeter et al> <Others>
< 5 -.021 -.033
5+ -.021 n/a
-
TucsonRoyal I would like to see the age vice experience. Scoring is less in the post season, mainly do to colder weather. Players that rely on hitting, younger players, will have less hits (e.g. a flyout that would have been a HR in warmer weather) while your older veterans will still be taking their walks, thereby keeping their wOBA up.
David G. checks in again on whether experience matters in the post-season
THE BOOK--Playing The Percentages In Baseball —
Here is the article on THT web site:
And here is what I wrote in the comments section:
Now I’m starting to get convinved, David! Maybe.
I don’t know that you are doing “pure cherrypicking.” In fact, I am pretty certain that you are not. If you had used 4 years or 3 years or 6 or 7 years as your cutoff point, and you got a completely different result, then I would call that “pure cherrypicking.” But, without doing the math myself, my guess is that any boundary you choose ...
Thursday Applesauce
Amazin' Avenue —
... At The Hardball Times, David Gassko follows up on last week's article about whether experience makes any difference to playoff success. ...
