Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Baseball Newsstand > Baseball Primer Newsblog > Discussion
Baseball Primer Newsblog
— The Best News Links from the Baseball Newsstand

Wednesday, December 02, 2009

Huffington Post: Bradbury: Is Your Favorite Free Agent Over the Hill?

If you follow sabermetrics, you’re probably familiar with the notion that the peak age for baseball players is 27. If this is the case, then the best players are several years past their prime. What does this say about how they’ll perform during the long-run contracts that top free agents typically sign? For example, last year’s top free-agent pitcher CC Sabathia will be seven years past his hypothetical prime once his contract expires when he’s 34.

There is no doubt that age will sap players of their ability over the course of most free-agent contracts; however, the decline isn’t as pronounced as many people believe. In a recent study, I examined the playing careers of many players over a span of 86 seasons to see how players aged. After controlling for a multitude of factors that might affect player performance I found players peaked around 29—-and this holds for hitters and pitchers. And I’ve proposed some explanations as to why past studies have found lower peak ages.

More important than the later peak is the fact that the rate of rise and decline is not all that pronounced. Though old players may not be what they once were, the evidence indicates they can still be valuable. Players are who they are, and will perform slightly better and worse than their peaks from their late-20s through early-30s. I find that players perform within two percent of their peaks from ages 26 to 32. Extending beyond this range, the dropoff isn’t particularly steep. According to my estimates, a hitter who has a .900 OPS at his peak would be expected to post around an .850 OPS at 35; a pitcher with a peak 3.5 ERA is expected to post around a 3.75 ERA at 35.

Thanks to Acres.

Repoz Posted: December 02, 2009 at 06:21 PM | 33 comment(s) Login to Bookmark
  Tags: sabermetrics

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

   1. JMPH Posted: December 02, 2009 at 06:32 PM (#3401230)
If you follow sabermetrics, you’re probably familiar with the notion that the peak age for baseball players is 27.

I think if you follow baseball at all, you're probably familiar with 27ish as a peak age.
   2. Harmon "Thread Killer" Microbrew Posted: December 02, 2009 at 06:39 PM (#3401239)
Is Your Favorite Free Agent Over the Hill?


Considering that my favourite free agent is Lloyd Moseby, all signs point to "Yes".
   3. Shooty Survived the Shutdown of '14! Posted: December 02, 2009 at 06:43 PM (#3401242)
Considering that my favourite free agent is Lloyd Moseby, all signs point to "Yes".

Dammit. I was going to make this joke except with Dwayne Murphy.
   4. Harmon "Thread Killer" Microbrew Posted: December 02, 2009 at 06:59 PM (#3401273)
Dear Shooty,

If you feel strongly about wanting that joke, I will remove my comment and you can post yours.

This will be our secret.

Let me know.

Your friend,

Harmon
   5. Shooty Survived the Shutdown of '14! Posted: December 02, 2009 at 07:05 PM (#3401279)
Sweet. Moseby is a poor man's Dwayne Murphy, anyway.
   6. AROM Posted: December 02, 2009 at 07:21 PM (#3401300)
According to my estimates, a hitter who has a .900 OPS at his peak would be expected to post around an .850 OPS


This is pretty bad. I'm fairly sure I know how he looked at this, and that is not an honest interpretation of the results. I believe he looked at players who were active at age 35 and at 27 or 29 or whatever his peak is. And he's got it backwards. Knowing that a player has an .850 OPS at age 35, he probably was a .900 player at age 27.

But if you're looking at age 27 here's what I find:
All players with OPS between .875 and .925, from 1993 to 2001, who were 27. (start of modern offensive era on one end, and the last year that a 27 year old could have and also have an age 35 season on record. B-Ref PI is my source. 18 players qualify, average OPS is .900

One player maintained a near .900 OPS, Jeff Bagwell. Another, Sammy Sosa, was at .849.

Six were between .770 and .804 (Ordonez, Karros, T Martinez, M Williams, M Sweeney, V Castilla).

Four were between .695 and .741 (Ventura, Catalonotto, John Valentin, I Rodriguez). There's Tim Salmon (628), Mike Lieberthal (.540, 82 PA). Juan Gonzalez almost made it through a single plate appearance before being done for good (but not quite, hurt running to first). Dean Palmer, Todd Hundley, and Trot Nixon were out of baseball.

The weighted average on PA of this group is .779, simple average of the ones who were still playing was .749, and the median (Robin Ventura) was .741

So given that a player has a .900 OPS at age 27, it is more likely that he's out of baseball entirely at 35 than still posting an .850 OPS or greater. And the median expectation of the guy is Robin Ventura 2003: 242/340/401, 392 AB, 14 HR, 55 RBI.
   7. Tango Posted: December 02, 2009 at 07:37 PM (#3401321)
If AROM/6 can be the last post in this thread, it would be perfect.
   8. SoSH U at work Posted: December 02, 2009 at 07:42 PM (#3401325)
If AROM/6 can be the last post in this thread, it would be perfect.


Or at least next to last.

Damn it.
   9. GuyM Posted: December 02, 2009 at 08:16 PM (#3401376)
Rally: could you give us a comparison of these players' RAR and WAR at both ages? My guess is that these declines are even sharper, as the 35-yr-olds tend to play less and to play less demanding defensive positions.
   10. JPWF13 Posted: December 02, 2009 at 08:47 PM (#3401425)
But if you're looking at age 27 here's what I find:



Aha! But peak is at 29!
Anyway, I did it for 1993-2003, all players age 29, 400+ PAs, OPS between .875 and .925, there were 18

Simple average age 29: .902
weighted average age 29: .902

4 were out of base ball at age 35
The simple average of the remaining 14*: .788
The weighted average of the remaining 14: .792

The median of the age 35 survivors was between Hal Morris and Cliff Floyd.

According to my estimates, a hitter who has a .900 OPS at his peak would be expected to post around an .850 OPS

I'm at a loss as to how he could have possibly arrived at that. (I suppose if your "peak" year was before 1993, and your age 35 season was after 1993...)



*I fudged a bit, Klesko didn't play at age 35, but he played at age 34 % 36, so I averaged those two ages together.
   11. AROM Posted: December 02, 2009 at 08:59 PM (#3401439)
Rally: could you give us a comparison of these players' RAR and WAR at both ages? My guess is that these declines are even sharper, as the 35-yr-olds tend to play less and to play less demanding defensive positions.


Didn't save it, or look at WAR, but all the players who met the criteria are listed in my post if you want to check. My assumption is that you are 100% correct, they are declining in defensive value as well as offensive.

I'm at a loss as to how he could have possibly arrived at that.


I think it is this simple: Find all players who have an .850 OPS at age 35, look back and you'll find they averaged about .900 at their peak. My guess is if i try that I'll be able to reproduce his results. His mistake is thinking that tells you anything about how we should expect an average 27 year old to do at age 35. He's got it all backwards.
   12. GuyM Posted: December 02, 2009 at 09:13 PM (#3401451)
I think it is this simple: Find all players who have an .850 OPS at age 35, look back and you'll find they averaged about .900 at their peak.

Almost. His paper looks at players who amassed 5000+ PAs from age 24 to age 35, and fits a quadratic equation to the players' seasons controlling for their career mean OPS, year/league, and other factors (and using only seasons with 300+ PAs). He finds a very flat aging curve overall, but some specific skills have distinct curves.

But essentially your intuition is right: he is really only looking at players who played well into their mid-30s, which guarantees a flat aging curve. And even the guys who do fall off a cliff in their early 30s and still make the 5000 PA cutoff don't have much impact on the curve, because most of their deline years are hidden from sight since they don't play at all or have <300 PAs.
   13. DKDC Posted: December 02, 2009 at 09:15 PM (#3401457)
A problem with the methodology in #6 is that you are qualifying players for this group based on a single season.

Mike Lieberthal was never a near .900 OPS true talent player. Even after his fluke age 27 year I doubt any projection system would have him anywhere near that high.
   14. JPWF13 Posted: December 02, 2009 at 09:20 PM (#3401469)
I think it is this simple: Find all players who have an .850 OPS at age 35, look back and you'll find they averaged about .900 at their peak. My guess is if i try that I'll be able to reproduce his results.


1975-92, just 12 batters averaged between .825-.875 at age 35:
Rk      Player      OPS      PA      Year      Age      Tm      Lg
1     Carlton Fisk     .874     545     1983     35     CHW     AL
2     Cliff Johnson     .862     484     1983     35     TOR     AL
3     Pete Rose     .854     759     1976     35     CIN     NL
4     Paul Molitor     .851     700     1992     35     MIL     AL
5     Tom Paciorek     .851     422     1982     35     CHW     AL
6     Jose Cruz     .848     664     1983     35     HOU     NL
7     Lou Whitaker     .847     544     1992     35     DET     AL
8     Brian Downing     .841     631     1986     35     CAL     AL
9     Jack Clark     .840     585     1991     35     BOS     AL
10     Gary Matthews     .839     432     1986     35     CHC     NL
11     Don Baylor     .830     558     1984     35     NYY     AL
12     Dave Kingman     .826     613     1984     35     OAK     AL 


Why yes, if you look at their age 27-29 peak year, it only averages around .875!!!!
In fact a surprising number of these guys didn't "peak" at age 27-29 at all, many were injured, it is a terribly unrepresentative list actually- plus by looking at guys who hit .850 at age 35, you are almost begging to find atypical spikes in player careers, or guys who peaked late.
   15. JPWF13 Posted: December 02, 2009 at 09:23 PM (#3401473)
OK, these guys, 1993-2001, averaged between .875 and .925 over their age 27-29 seasons:

Rk      Player      OPS      PA      To      From      Age      G
1     Carl Everett     .924     1615     1998     2000     27
-29     393
2     John Olerud     .921     1764     1996     1998     27
-29     439
3     Chris Hoiles     .921     908     1993     1994     28
-29     225
4     Tim Salmon     .912     1942     1996     1998     27
-29     449
5     Shawn Green     .905     1415     2000     2001     27
-28     323
6     Vinny Castilla     .902     1912     1995     1997     27
-29     458
7     Rusty Greer     .900     1998     1996     1998     27
-29     451
8     Reggie Sanders     .899     1259     1995     1997     27
-29     300
9     Sammy Sosa     .898     1957     1996     1998     27
-29     445
10     Tino Martinez     .898     1949     1995     1997     27
-29     454
11     Reggie Jefferson     .895     1161     1996     1998     27
-29     320
12     David Justice     .886     1585     1993     1995     27
-29     381
13     Jay Buhner     .886     1111     1993     1994     28
-29     259
14     Ryan Klesko     .884     1546     1998     2000     27
-29     407
15     Roberto Alomar     .879     1745     1995     1997     27
-29     395
16     Jeffrey Hammonds     .878     1110     1998     2000     27
-29     334
17     John Valentin     .878     1582     1994     1996     27
-29     350
18     Todd Hundley     .877     1274     1996     1998     27
-29     338
19     Jeromy Burnitz     .876     1507     1996     1998     27
-29     408 


Any guesses as to how those guys did in their age 35 seasons?
A whole lot worse than .850
   16. Gaelan Posted: December 02, 2009 at 09:27 PM (#3401477)
I'm shocked to discover that Bradbury was wrong about something.
   17. DKDC Posted: December 02, 2009 at 09:47 PM (#3401506)
I just did the same thing as #15, with a 1500 PA threshold, and for ages 26 – 29.

I extended the OPS range from 875 to 950 OPS to get to an average of 900 OPS.

16 players, average OPS of .898.

Their average OPS from ages 33 to 35 was 0.822, for an average of .076 drop.

If you exclude the two catchers (who we all know decline faster), it’s a 0.63 drop.

26-29 33-35 Diff Player
0.877 0.995 0.118 Jim Edmonds
0.930 1.031 0.101 Larry Walker
0.884 0.922 0.038 Sammy Sosa
0.886 0.882 -0.004 David Justice
0.876 0.856 -0.020 Reggie Sanders
0.878 0.798 -0.080 Jeromy Burnitz
0.891 0.802 -0.089 John Olerud
0.887 0.798 -0.089 Tino Martinez
0.936 0.833 -0.103 Bernie Williams
0.940 0.824 -0.116 Tim Salmon
0.877 0.733 -0.144 Rusty Greer
0.927 0.779 -0.148 Shawn Green
0.875 0.724 -0.151 Todd Hundley
0.878 0.714 -0.164 Carl Everett
0.899 0.721 -0.178 Vinny Castilla
0.919 0.740 -0.179 Ivan Rodriguez
   18. Yeaarrgghhhh Posted: December 02, 2009 at 09:58 PM (#3401518)
DKDC -- can you do the same thing using OPS+ instead? maybe 125-150 for the peak years.
   19. AROM Posted: December 02, 2009 at 10:10 PM (#3401538)
I suppose Bradbury could be right on this "a hitter who has a .900 OPS at his peak would be expected to post around an .850 OPS at 35" if he gave you the qualification that the player at 35 would still be playing, and playing everyday. But if you are deciding whether to pull the trigger on an 8 year deal for such a player, you need to know what are the odds he'll still be playing at all, and if so what are the odds he's still able to play regularly.
   20. JPWF13 Posted: December 02, 2009 at 10:25 PM (#3401559)
If you exclude the two catchers (who we all know decline faster), it’s a 0.63 drop.


But the median is still .89, plus Y is right, we should be using EQA or OPS+ or something, a .900 in Coors isn't the same as a.900 in Seattle
   21. DKDC Posted: December 02, 2009 at 10:37 PM (#3401570)
Yes, these are all quick and dirty analyses that could be improved if someone was willing to put some time into it. A bigger sample and park adjustments would certainly help.

My WAG is that a rigorous study would put the OPS drop somewhere between Bradbury's 50 and AROM's 150.
   22. JPWF13 Posted: December 02, 2009 at 10:44 PM (#3401576)
Ok ages 26-29, OPS+ 120-125, 1500+ PAs: (1987-2003):
Player    OPS+
Tony Clark    125
Tino Martinez    125
Jay Buhner    125
Roberto Alomar    124
Robin Ventura    124
Barry Larkin    124
Trot Nixon    123
Rondell White    123
Carl Everett    123
Ryan Klesko    123
Jim Edmonds    123
Derek Jeter    122
Rusty Greer    122
Chuck Knoblauch    122
Dave Magadan    122
Jose Vidro    121
Hal Morris    121
Lenny Dykstra    121
Wally Joyner    121
Bobby Higginson    120
Javy Lopez    120
Dave Nilsson    120
Moises Alou    120 

7 were out of baseball by age 35
the 16 remaining averaged 102 (simple average) in their age 35 seasons

4 exceeded their age 26-29 average in their age 35 season:
Jim Edmonds (he really did peak in his early to mid 30s)
Wally Joyner (he had a BABIP driven spike)
Derek Jeter (his age 34-35 average would be below)
Jay Buhner (by 1, and just a 126 in only 400 PAs, sandwiched by a 105 and a 100)
   23. JPWF13 Posted: December 02, 2009 at 10:46 PM (#3401578)
My WAG is that a rigorous study would put the OPS drop somewhere between Bradbury's 50 and AROM's 150.


what do you do with the guys who aren't playing at all at the later age?

BTW in the Article there's a link to JC's study- anyone wanna pay $30 for it?
   24. 1k5v3L Posted: December 02, 2009 at 10:56 PM (#3401590)
Eric Byrnes will bounce back, dammit! He's my dark horse to win next year's NL MVP.
Jeff Moorad thinks the world of Eric, and that's usually good enough for me.
   25. Bhaakon Posted: December 03, 2009 at 08:31 AM (#3401873)
what do you do with the guys who aren't playing at all at the later age?


You could supply two numbers: odds that they'd still be in baseball, and a projection if they are. I suppose that you could just fill their empty seasons with whatever replacement level was that season and continue running averages, but I don't think that would be a particularly accurate representation of reality (since the ones still playing would beat the projection by a significant margin, and the ones out of baseball would fall hugely short).
   26. Walt Davis Posted: December 03, 2009 at 09:17 AM (#3401881)
If you used WAR or even BRAA or something, you could pretty easily control for dropout rate:

E(WAR at age 35) = p(out of bb by 35)*0 + (1-p)*E(WAR | play at 35)

Since the first term disappears, that's the same as the total age-35 WAR of the surviving players divided by the full sample size. I suppose technically you could do that for OPS but interpreting the answer would be a challenge.

It gets a bit more complicated to, say, look over the life of a 6-year contract to a guy entering his age 30 season (esp since a guy could miss a season at age 30 then return at 31), but the same basic technique would work. Of course to apply that to an individual player is making the assumption that everyone in the group is the same in terms of risk of not playing at 35 and likely age-35 performance. But close enough for a start.
   27. Dr Stankus and the Semicolons Posted: December 03, 2009 at 03:54 PM (#3402008)
Yes, these are all quick and dirty analyses that could be improved if someone was willing to put some time into it. A bigger sample and park adjustments would certainly help.


Perhaps by sending it through the academic systems, as Bradbury does...
   28. GGC don't think it can get longer than a novella Posted: December 03, 2009 at 04:11 PM (#3402040)
What do they call this in murtual fund analysis? Something like Survivorship Bias, IIRC.
   29. snapper (history's 42nd greatest monster) Posted: December 03, 2009 at 04:36 PM (#3402063)
What do they call this in murtual fund analysis? Something like Survivorship Bias, IIRC.

Yes. Just like your life expectancy gets higher the older you get.

The futures fund industry is notorious for using/abusing this. A firm will start 5 funds with different strategies. After 5 years one will have a good track record, and they'll tout that one to investors, while letting the other 4 die.
   30. AROM Posted: December 03, 2009 at 05:23 PM (#3402126)
I've read through the article and some linked entries on Bradbury's blog. I won't anything because for some reason his site won't allow copy/paste. So to summarize:

We know he's only looking at players with long careers: 5000 PA and 10 seasons. He's responded to critiques of the selected sampling with some bizarre reasoning. He makes a distinction between aging and injuries: More players will peak at 27 (the mode method, look at what age most commonly has a player's best year) because some get hurt and don't have an opportunity to play at later ages.

On not using short career players he responds to the case of Marcus Giles by claiming that whatever caused his decline is a mystery (which is certainly true) but it's not aging.

Well, if that's the way you want to play it, fine. If I know that a certain 27 to 29 year old will still be playing at 35, I know he won't get hurt, and I know he won't turn out like Marcus Giles, then we should expect only a small dropoff in his hitting stats.

But my point is, you don't know any of that. And it's all relevant, risk that must be accounted for, by a team considering signing a 29 year old free agent.
   31. JPWF13 Posted: December 03, 2009 at 05:46 PM (#3402160)
On not using short career players he responds to the case of Marcus Giles by claiming that whatever caused his decline is a mystery (which is certainly true) but it's not aging.


But who the eff cares if it's not "aging".

If you are going to sign a 29 year old to a 6 contract, you'd still should know how many 29 year olds mysteriously evaporate due to "non aging" reasons.

Maybe, just maybe, a 29 year old OPSing .900, who keeps in shape, doesn't get sick/ run over by a car, has an even money shot of OPSing .850 when he's 35.

But when you sign that 29 year old YOU DO NOT KNOW he is:
1: not going to get hurt
2: not going to contract aids/swine flu/ebola etc;
3: not going to take up recreational eating;
4: not going to have a premature midlife crisis;
5: not going to get three fingers crushed in a limo door;
6: etcetera

Crap happens, personally I see crap happening as part of aging/experience
   32. Gaelan Posted: December 03, 2009 at 06:02 PM (#3402181)
This is fun and all but at some point we have to realize that Bradbury is incapable of thought and not bother. Here is the new Sabermetric golden rule. When I doubt the answer is the exact opposite of whatever Bradbury thinks.
   33. GGC don't think it can get longer than a novella Posted: December 03, 2009 at 06:26 PM (#3402219)
How does he feel about DIPS, Gaelan?

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
Forsch 10 From Navarone (Dayn)
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Hot Topics

NewsblogOT: Politics, August 2014: DNC criticizes Christie’s economic record with baseball video
(6027 - 11:29pm, Aug 28)
Last: Jolly Old St. Nick Is A Jolly Old St. Crip

NewsblogJack White, Eddie Vedder, and Paul Simon take in a Seattle Mariners game
(146 - 11:28pm, Aug 28)
Last: PreservedFish

NewsblogFG: A Good Reason To Watch Yusmeiro Petit Pitch
(15 - 11:25pm, Aug 28)
Last: SoSHially Unacceptable

NewsblogRingolsby: Pete Incaviglia paying dues as independent league manager
(5 - 11:23pm, Aug 28)
Last: SoSHially Unacceptable

NewsblogDavid Justice Says Put Barry Bonds in Baseball Hall of Fame Despite Steroid Use Late In Career
(140 - 11:15pm, Aug 28)
Last: Rennie's Tenet

NewsblogOMNICHATTER 8-28-2014
(33 - 10:55pm, Aug 28)
Last: Los Angeles El Hombre of Anaheim

NewsblogBrooklyn Cyclones, Nickelodeon to host '90s night
(30 - 10:49pm, Aug 28)
Last: bigglou115

NewsblogSimmons' run-saving stop
(34 - 10:48pm, Aug 28)
Last: PreservedFish

NewsblogOT: Monthly NBA Thread - August 2014
(363 - 10:48pm, Aug 28)
Last: Crispix reaches boiling point with lackluster play

NewsblogPosnanski: Alex Gordon and the M-V-P chants
(17 - 10:30pm, Aug 28)
Last: Kiko Sakata

NewsblogPrimer Dugout (and link of the day) 8-28-2014
(18 - 10:21pm, Aug 28)
Last: The District Attorney

NewsblogPadres, City of San Diego pay homage to Selig
(10 - 10:13pm, Aug 28)
Last: Cargo Cultist

NewsblogFangraphs: Sullivan: Anatomy of a Disaster.
(8 - 10:07pm, Aug 28)
Last: Buck Coats

NewsblogBusiness Week: Houston Astros' Jeff Luhnow Lets Data Reign
(2 - 10:05pm, Aug 28)
Last: Cargo Cultist

NewsblogOT: The Soccer Thread August, 2014
(769 - 9:40pm, Aug 28)
Last: Darkness and the howling fantods

Page rendered in 0.4369 seconds
52 querie(s) executed