Do fast players age more gracefully than slow players?
All things being equal, what kind of star hitter ages more gracefully – a fast player, or a slow player?
This Tangotiger study, linked to by Tango on his blog, finds out.
I love this study. Actually, I love most of the stuff Tango comes up with, but this is one of my favorite kinds of studies. It's simple, it doesn't rely on any fancy statistical methodology, and the results are easy to understand and hard to dispute.
Tango found the twenty "best-hitting speedsters" up to age 30, from players born 1941-1960. Then, for each of the twenty, he used some version of similarity scores to find the most comparable slowster – that is, a player similar in every category except speed (3B and SB/CS, presumably – Tango doesn’t give us his actual similarity formula).
Tango observes that part of the fun is seeing who matches up with whom; my favorite, for some reason, is Al Oliver as a slow Ralph Garr.
Before age 30, the two groups are about the same (per 600 PA):
G AB R H 2B 3B HR RBI BB
Fast 143 538 87 154 25 7 12 59 54
Slow 149 535 72 152 27 3 15 73 56
And after 30? Again, almost the same:
G AB R H 2B 3B HR RBI BB
Fast 154 528 83 149 26 5 14 67 63 (total 3307 PA)
Slow 158 533 70 149 26 3 15 75 59 (total 2938 PA)
The conclusion: skilled fast players age at about the same rate as skilled slow players, but manage to hang on for an extra 12% more career playing time.
This result appears to contradict a similar study from Bill James. In the "Rookies" section in the 1987 Abstract (p. 67), Bill James looked at 40 pairs of players, selected similarly from rookies throughout baseball history. The results were much more dramatic: 43% more major league games for the fast players. James didn't give full stats for the two groups, but he did say that the fast players had only 38% more home runs in 43% more games, which means that the fast players hit for less future power than the speedsters.
Why do the two studies show such a drastic difference in the effect of speed on career length? I can think of a couple of reasons, theoretically possible but probably not true:
The James study compared all kinds of young players after only one season; Tango's study compared only older star players with established records. Maybe players slow down most between 20 and 30. Most guys who are slow at 24 are going to be too slow to play by 29; but the ones who stay in shape are as slow at 30 as they're ever going to get, and can stick around longer.
Or maybe fast young guys tend to rise faster than slow guys up to age 27, but decline at the same rate after. That would explain why the two groups are different from age 24-40, but the selectively sampled older players are similar from age 30-40.
Anyway, I bet the correct explanation is something along these lines.
Labels: baseball
3 Comments:
interesting piece on aging, but there is no significant between this matching & james'.
the competing study only includes players who stayed healthy enough to be booted out of the majors because of performance.
there is a large element of self-proving here.
the interesting bit is that slow players--not shown in this study--have a much steeper slide on the right hand of their careers than do fast players.
slow players' future value is much more easily predictable than quick, low walk, low power, poor throwing centre fielders.
The two studies are different, and so, I wouldn't necessarily want to compare the two. As long as the reader is aware of the selection criteria that goes into my study (quality players through age 30), I wouldn't want them to try to extrapolate beyond that.
At the same time, it sets the stage for others to pick up the mantle and try other aging studies.
Along with the catchers study, this is my favorite study.
The results of this particular study ... shows that speedsters and leadfooters show no difference in their aging patterns.
There's a significant difference between 'aging patterns' and 'what is the average between his peak and retirement?'. If both speedsters and leadfooters play until they're worth some marginal quantity and both age linearly, they will have the same average. That is to say int(1-x,x=0..1)/1=int(1-x/2,x=0..2)/2
[Maple formatted integrals]
A player who plays one season after turning 30 and retires may have the same averages (for games where age>30) as a player who plays until 40 (for games where age>30), but clearly the 40 year old aging patters are different than the 31 year old.
In other words the fact that there were more PA for the speedsters, even if not statistically significant, warrants additional study before making a final conclusion.
All this tells me is that good player's past performance can predict future performance quite well, independent of type of player (speedsters or leadfooters).
Post a Comment
<< Home