Can We Use Bonds' Career WAR Arc to Identify PED Users? In a word, no.

(Click to enlarge)
Justin’s post on Gary Sheffield the other day prompted a brief discussion about Barry Bonds and his famous "double peak" arc in the comments section. The question arose whether we could use Bonds’ arc as a method for locating players whose performance was assisted by banned substances. Since there is a preponderance of evidence that Bonds used various steroidal and anabolic substances, can’t we just chalk up his "double peak" to the drugs and look for other players that exhibit the same arc?
It’s not out of the question, but it’s far from a solid methodology for a number of reasons. The first one I’ll address is that, as far as elite players goes, no one has ever exhibited the kind of double peak that Bonds did. Bonds is unique not only because he exhibited a second, sustained peak in performance late in his career, but because that second peak was better than his first.
To find a group of comparable players I used Baseball-Reference’s Play Index to locate players age 34 or older that managed a WAR of >=8 in a given season. As you can imagine, the list was pretty small. There have been only 21 seasons that match that description.
Barry Bonds heads the list with five seasons, Babe Ruth had four, Honus Wagner managed three, Willie Mays had two, and seven other players managed just one (Ted Williams, Eddie Stanky, Tris Speaker, Nap Lajoie, Lou Gehrig, Jim Edmonds, and Hank Aaron).
Given that only four players managed more than one 8+ WAR season after age 34 it made the task of comparing arcs pretty easy. I plotted the WAR by age for Bonds, Ruth, Wagner, and Mays. The data confirms that, while some players have gone on to have spectacular seasons after 34, no one has performed like Bonds, making any diagnostic instrument based on his career WAR arc suspect.
(Fair warning--my WAR Arc's are not as pretty as Justin's.)

(Click to enlarge)
Ruth’s best season was at age 28 (an amazing 14.7 WAR). He had his 3rd and 5th best seasons at age 32 and 36, but his general arc seems to suggest a single peak at age 28 with gradual decline beginning around 32.

(Click to enlarge)
Willie Mays had his best season at age 34, posting a WAR of 11 capping off four straight seasons of 10+ WAR. After his first two seasons (and a year of military service), Mays had his first 10 WAR season at 23 and then averaged 9.5 WAR from 23 to 34 only dropping below 7 once and 8 twice. Mays is odd in that he never really had an observable peak, sustaining excellence straight form 23 to his peak year at age 34. He certainly did not Bonds’ "double peak" arc, as his performance starts to permanently decline after that career year.

(Click to enlarge)
What about everyone’s favorite rare baseball card, Honus Wagner? He was a beast, peaking at age 34 while posting a WAR of 11.6. Looking at Wagner’s arc, however, it looks quite normal—sustained excellence from age 25 to 34 with a steady, productive decline beginning immediately after.
The point of this exercise was to show that Bonds is an outlier, plain and simple. Might we be able to develop a tool to identify PED users? Sure, it’s possible (although riddled with methodological hurdles that I won't list here). But we can’t use Bonds’s arc as a road map, even for elite players.
There simply hasn’t been anyone, PED user or not, like him.
10 comments
|
2 recs |
Do you like this story?
Comments
1) The double peak is function of a few down years from his age 29-30 seasons and age 34 season. Wasn’t he injured those years? In other words, he didn’t experience a sustained drop due because he lost his skills and then bounced back. It seems more likely that he got injured, had some a few down years as a result, and then continued playing to his skill level. If he would have stayed healthy, the his career arc might look more like a normal curve instead of a double curve.
2) I’m not sure why we would use any one case to identify a trend in anything.
3) Instead of looking for double peaks, I think it may be interesting to look for sustained performance relative to past performance late into a player’s career.
Folksy literate type.
I agree
I think this looks more like a single arc with some variation due to injuries (and probably other variables as well) — there’s nothing at all like a sustained, significant drop in real value there, at least through Bonds’ 30s.
But then, for my part, I still consider the idea that steroids, etc. produce a significant increase in hitting ability to be merely inferred, not proven. Correlation does not equal causation, after all, and I’m not really all that impressed by even the demonstrated correlation to this point.
by The Ancient Mariner on Feb 24, 2011 2:27 PM EST up reply actions
I think this is semantic to some degree.
Call it an excellence platue if you like. Either way, Bonds was elite from 24-35 and then experienced another level of performance. This pattern is like nothing we’ve ever seen, so I think the general point still holds.
As for why we would think about creating a “cheater alert” based on an N of 1, I don’t believe it makes much sense but it had been suggested, which was why I did the research.
Writer at Beyond the Box Score and tortured Mets fan (is there any other kind?)
Either way, Bonds was elite from 24-35 and then experienced another level of performance. This pattern is like nothing we’ve ever seen, so I think the general point still holds.
Right, so instead of looking for a double curve, it may be more interesting to look for an excellence plateau in relation to past performance. In other words, instead of only looking at elite players, it would be interesting to see if 2 win players remain 2 win player longer than what past findings suggest.
Folksy literate type.
I don't see the double peak for Bonds at all.
Bonds’ age 29 season was the strike-shortened 1994, when he posted 6.4 WAR (bb-r numbers) in 112 games. He was well on his way to a solid 9 WAR season that year. In ’95 he posted an OPS+ of 168 and led the league in batting wins. In ’99 he was injured, played 102 games, and still managed 4.0 WAR and a 155 OPS+. Babe Ruth had a worse year in 98 games in his age 30 season. Is he a double peaker?
Proud father of Barry Bonds.
As I said above
No player has ever peaked the way Bonds has at age 36. If we look at the general arc of peak performance it starts much earlier. Bonds wasn’t building toward anything from 24-35, he achieved sustained excellence. He then created another peak from 36-39.
If you want to say post-36 peak, fine, but I think getting stuck on the whole double peak, single peak thing misses the point.
Re: Ruth, no. Secondary peaks, for me, suggest not simply high performance but establishing a new best. Ruth’s peak was well before age 30.
Writer at Beyond the Box Score and tortured Mets fan (is there any other kind?)
Mays
averaged 9.5 WAR from 23 to 34 only dropping below 7 once and 8 twice
Hot damn.
On Twitter: @baseballtwit
yeah, that really stood out for me--incredible.
Writer at Beyond the Box Score and tortured Mets fan (is there any other kind?)
I was looking at this graph
without looking at the scale, and I was like, “wow, look at all those 8 WAR seasons!”
/facepalm
they were more like 10 WAR seasons. Goodness gracious
San Francisco Giants: 2010 World Series Champions
Buster Posey: 100% ballplayer, 0% bullshit
by free f.p. #14 on Feb 24, 2011 5:26 PM EST up reply actions

by 





























