Aging patterns (September 23, 2003)
I did this a few years ago, and there are adjustments I should make, but I never got around to it. In any case, I made the following statement
Other than the triples as I noted, the historical averages much up very well with the recent period. While today's ballplayers may be better, and playing longer, the "curve" of their aging is the same. There is no age bias with today's regimen of training and medication. It affects all age groups the same.
Bill James reported the same thing in one of his 2 last books. Rob Neyer reiterated James' finding in his last article, which is why I thought about posting this here.
--posted by TangoTiger at 05:15 PM EDT
Posted 6:09 p.m.,
September 23, 2003
(#1) -
Admiral Ackbar
It's a trap!
Posted 6:16 p.m.,
September 23, 2003
(#2) -
studes
(homepage)
Thanks, Tango. Has anyone ever done a similar study for pitchers? Seems to me that pitchers age differently than hitters, but there's also probably a lot more data noise. Are you aware of any similar pitcher studies?
Posted 7:24 p.m.,
September 23, 2003
(#3) -
studes
(homepage)
BTW, after staring at this chart a bit, what does this mean about players being able to learn to take a walk? Doesn't Beane propose, in Moneyball, that players can't be taught plate discipline? And doesn't this seem to run counter to that argument?
Posted 7:32 p.m.,
September 23, 2003
(#4) -
studes
(homepage)
Okay, last post for now. Why the heck does the $H go down from year one??? That floors me. Is it due to speed? I could understand sample bias issues at the age extremes, but that is just weird.
Other observation: the walk rate and hit rate trend lines intersect between ages 27 and 28, right around the $LW peak. Don't know if that means anything, but it is interesting.
Posted 11:42 p.m.,
September 23, 2003
(#5) -
Tangotiger
Yes, I did the same for pitchers, and they peaked at age 24.
The problem is that I didn't handle "regression towards the mean", the single most important thing to understand if you're going to do these things. I was unaware of the concept then, and will eventually redo these charts for pitchers and hitters, as well as taking care of park and era adjustments.
The strike zone IS a learned skill, but it is based on experience and intelligence. Even things like power IS a learned skill. You may physically mature at age 23 or so, but hitting HR or hitting with power also requires experience and maturity. It's not like it's a T-ball league.
Posted 12:10 a.m.,
September 24, 2003
(#6) -
studes
(homepage)
Ah. Not T-ball. I'll try and remember that.
Posted 12:26 a.m.,
September 24, 2003
(#7) -
Tangotiger
I wasn't trying to be facetious or anything. It's basically bs when someone says that you can "learn" something, and I don't care if it's Beane or Bill James that says it.
What they are probably trying to say is that there is a certain learning pattern that is being followed, and you won't be able to change that pattern too much. That is, whether you are Frank Thomas or Alfredo Griffin, you may start off at different levels, but you will both follow the curves that are based on your skillset, and that those curves exist because of the experience that you will get.
Usually, the right answer is the easy answer, and I believe this is the case here. I haven't proved it, but neither has the opposite been proven by their believers either.
Posted 12:26 a.m.,
September 24, 2003
(#8) -
Tangotiger
"you can " = "you can't"
(That changes the whole meaning!)
Posted 12:38 a.m.,
September 24, 2003
(#9) -
studes
(homepage)
It's an interesting discussion from a GM's point of view. How much you decide to invest in a young developing player depends on your perception of how well he can learn certain skills. I'm guessing development people make this sort of calculation every day: ability to learn vs. potential if learning occurs.
Skills can be learned, but it's very hard to change attitudes and nearly impossible to change personalities. Maybe a lot of baseball people believe plate discipline is an attitude, not a skill.
Sorry about the ramble. I'm still not sure what to make of the BABIP decline. Is that warped by lack of regression to the mean?
Posted 7:40 a.m.,
September 24, 2003
(#10) -
Tangotiger
It's either that, or that players change their approach to hitting as they age. Once you have confidence to be patient and hit HR, why settle just for a BIP?
Posted 9:43 a.m.,
September 24, 2003
(#11) -
David Smyth
What about loss of speed having a role in the BABIP decline?
Posted 10:06 a.m.,
September 24, 2003
(#12) -
tangotiger
Great call!
If I had the data, it would have been nice to break it down into $Hfly and $Hground.
I suppose I can redo the charts eventually, one set for "fast" runners and one for "slow" runners, and see what the aging patterns are. I'm sure PECOTA has done this kind of work already.
Posted 11:13 a.m.,
September 24, 2003
(#13) -
studes
(homepage)
Hey, I referred to speed in #4! I guess great minds think alike...
Nah. I'm just greedy for some credit. :)
Posted 1:21 p.m.,
September 24, 2003
(#14) -
David Smyth
Sorry for the rip-off, studes. It wasn't intentional.
Posted 2:08 p.m.,
September 24, 2003
(#15) -
studes
(homepage)
NP, David. I really don't mind at all. Just kidding.
Posted 6:21 p.m.,
September 25, 2003
(#16) -
bob mong
I tried to sorta replicate this using the Lahman db, Tango, and I got mostly the exact same figures, except:
I got $H peaking between 23 & 26 and not falling nearly as far; reaching a low of .88 at age 39 (you show, a low of .72 at age 39).
How did you get your numbers?
Posted 2:03 p.m.,
September 26, 2003
(#17) -
tangotiger
I probably messed up somewhere. Remember, I jsut did those real quick a few years ago. That's why it looks like that, instead of polished.
Posted 7:43 p.m.,
September 28, 2003
(#18) -
RossCW
Why the heck does the $H go down from year one??? That floors me. Is it due to speed? I could understand sample bias issues at the age extremes, but that is just weird.
I doubt there is a big drop off in speed from 22 to 23. You will notice more balls are going out of the park as they get older - that probably means fewer well hit balls in the park. There is also likely a sample bias that players who are speedy slap hitters get to the big leagues faster than lumbering power hitters.
Posted 9:43 p.m.,
September 28, 2003
(#19) -
Dylan Wright
So, to read this chart correctly. If I have a plyer who is 36 and I'm trying to estimate his 37 year stats:
Would I take the factor by itself or would I use the change between the age 36 factor and the age 37 factor.
IOW, are the factors absolute or relative?
Posted 9:55 p.m.,
September 28, 2003
(#20) -
Tangotiger
Suppose that HR (age 36)=30, HR factor (age 36) = .90, HR factor (age 37) = .87, you do:
30/.90*.87 = whatever-that-comes-out-to
Posted 10:10 p.m.,
September 28, 2003
(#21) -
Dylan Wright
I assume you mean (30/.90)*.87 which would be 29 total as opposed to 30/(.9*.87) which would be 38. I have been doing it (.87/.9)*30 which also gets you 29 HRs (but that's just Algebra)
Posted 10:17 a.m.,
September 29, 2003
(#22) -
tangotiger
What's cool about 30/.90 is that it gives you the player's "peak". That is, you would do 30/.90 x 1.00 to give you the HR at the peak age.
Of course, there is GREAT variability, so be careful with this train of thought.
Posted 12:47 p.m.,
September 29, 2003
(#23) -
FJM
Also remember that the ratios in the table are based on HR Rates, not the raw number of HR's. His number of AB's is likely to decline significantly as a player reaches his late 30's, even if his production rate remains fairly constant. Also, his strikeout rate will increase, giving him relatively fewer opportunities to make contact.
One thing I don't understand, TT: why do you subtract HR's from the denominator: HR/(AB-K-HR)? Also, why subtract K's from the denominator of the K-rate: K/(AB-K)?
One
Posted 12:57 p.m.,
September 29, 2003
(#24) -
tangotiger
That's how you can do chaining. I explained this in a long-deleted post at fanhome.
In essence, if you don't do things as ratios, you will get results that are mathematically incorrect.
That is, if you did aging rates for 1b/pa,2b/pa,3b/pa,hr/pa,bb/pa AND outs/pa, and redid this chart for ages 21 to 37, you will not be able to do it.
If you really want me to get into it, I see if I can dig something up. But, ratios, and not percentages, is how you need to do this.
Posted 12:05 p.m.,
September 30, 2003
(#25) -
bob mong
Since I built a whole spreadsheet last week using the chaining, I will give a stab at explaining:
Say you know a guy's AB, BB, 1B, 2B, 3B, HR, SO, and SB for his age 25 season.
To predict his age-26 season, given his plate appearances (AB+BB) in his age-26 season, you would:
Figure his previous age-25 $BB, figure out the new rate (multiply the old $BB by .82/.84, in this case), and then, to get age-26 ABs and age-26 BBs, solve this equation:
$BBnew = (BBnew)/(ABnew) = (BBnew)/(PAnew - BBnew)
rearranging...
$BBnew * (PAnew - BBnew) = BBnew
($BBnew * PAnew) - ($BBnew * BBnew) = BBnew
$BBnew * PAnew = BBnew + ($BBnew * BBnew)
$BBnew * PAnew = BBnew * (1 + $BBnew)
Which gives us (if I did everything correctly)...
BBnew = ($BBnew * PAnew)/(1 + $BBnew)
Everything on the right side we already know, so we can find his age-26 BB (denoted as BBnew). Since we know that PA = AB + BB, then we can find his age-26 AB as well.
Now, if you look at the next rate ($K), you can see that it only uses strikeouts and AB. Since we know his new AB total, we can similarly figure his new K total, and onward and upward, etc.
Did this make sense?
Tango, when you ran your numbers, did you adjust for league-context?
Posted 12:45 p.m.,
September 30, 2003
(#26) -
tangotiger
I did not adjust for park or league or anything. The only reason is because I did not do this for any purpose other than to get a general idea.
Posted 12:50 p.m.,
September 30, 2003
(#27) -
bob mong
Thanks, Tango.
Sorry to keep badgering you about this; I was just curious (mostly to know if my numbers are way off, indicating that I totally screwed up somewhere). :)
Posted 2:26 p.m.,
October 1, 2003
(#28) -
tangotiger
(homepage)
I promise I'll get to this again eventually bob!
At the above link is an email exchange I had with Mike Gimbel on the peak age of a player (something about the peak age being later than we think). This was posted in Jan at fanhome, but might be worthwhile for Primer readers who've never seen it. Mike was understandly tight-lipped about his thoughts, as he was trying to keep things proprietary and valuable to him.