Did Bumgarner and Shields Throw Too Many Pitches?

Madison Bumgarner pitched quite a bit this past season. Including the regular and post season, he threw a total of 4,074 pitches, which wasn’t even the season’s top total; James Shields bested him by throwing six more, for a total of 4,080 pitches in 2014, not including spring training. So with all of the pitches thrown this season (and one month less of rest), how should we expect these two to produce next season? Let’s look at some comparable pitchers.

To get the information, I looked at all pitchers from 1995 to 2013 who threw at least 4,000 pitches. I found thirty of them. Randy Johnson showed up on the list five times, and Justin Verlander was second with four appearances. In all, 21 different pitchers made the list which ranged from Justin Verlander throwing 4,004 in 2009 to Randy Johnson using over 4,700 (!) in 2001. Livan Hernandez in 2005 was the only pitcher to top 4,000 pitches in just the regular season. Finally, these pitchers were on the older side, with their average age being over 31-years-old.

So here are the pitchers and the seasons I will be comparing

A.J. Burnett (2009), Adam Wainwright (2013), Al Leiter (2000), Barry Zito (2003), C.J. Wilson (2011), CC Sabathia (2009), Chris Carpenter (2011), Cliff Lee (2009), Curt Schilling (1998,2001), John Smoltz (1996), Jon Lester (2013), Justin Verlander (2009,2011,2012,2013), Kerry Wood (2003), Kevin Brown (1998), Livan Hernandez (2005), Pedro Martinez (2004), Randy Johnson (1995,1999,2000,2001,2002), Roger Clemens (1996,2001), Roy Oswalt (2005), Russ Ortiz (2002), Tim Lincecum (2010)

For each of these pitchers, here is their combined 4,000-pitch regular season numbers (no post season values) and how they performed the next regular season. Additionally, I have included Marcel projection for the high pitch count season and the one after it.

Note: For pitchers, I like using the median values because one high ERA can skew the value up. With these pitchers, the average and median values are too far off, but I will use median values from now on.

    4,000 pitch season   Next Season  
    Marcel Projection Regular Season Marcel Projection Regular Season
Median ERA 3.58 3.21 3.36 3.27
  IP 190 238 200 217
  Age 31.5 31.5 32.5 32.5
Average ERA 3.48 3.15 3.39 3.38
  IP 189 235 199 204
  Age 31.6 31.6 32.6 32.6

Going into the 4,000-pitch season, the group of pitchers were supposed to have an ERA just under 3.50 and throw about 190 innings. For these 30 pitchers, they hit the trifecta.

1. They out performed their ERA by 33 points. Every season some pitchers will do better than projected, others worse. These were the ones who did better, which is why they were allowed to throw so many pitches to begin with.
2. They were healthy. A pitcher doesn’t reach 4,000 pitches by taking a month off with a dead arm.
3. The combination of the #1 and #2 probably boosted the pitcher’s production enough to help their team reach the postseason, where their workloads were extended even further.

So now we look at how they performed the next season. As expected, they performed worse. Comparing season one to season two stats, their ERA jumped by 23 points and they threw 23 less innings. Then look at the projections, the ERA value is off by 1 point and the innings projection missed by 5. The projections where about dead on as a group. Via regression and aging, the pitchers performed worse, but their production didn’t fall off a cliff. Justin Verlander‘s ERA underperformed his projections the most (projected 3.12 compared to 4.54 actual) while Roger Clemens was projected at 3.92 in 1997, but it ended up at 2.05. Only twice did a pitcher see his innings drop by more then 100 and each time it was Randy Johnson (1995 to 1996 and 2002 to 2003).

If a person wants a good idea of how James Shields and Madison Bumgarner will perform in 2015, after throwing 4,000 pitches in 2014, just look at their projections. They could under or over perform, but if history tells us anything, it’s that we should trust the projections.



Print This Post



Jeff writes for FanGraphs, The Hardball Times and Royals Review, as well as his own website, Baseball Heat Maps with his brother Darrell. In tandem with Bill Petti, he won the 2013 SABR Analytics Research Award for Contemporary Analysis. Follow him on Twitter @jeffwzimmerman.


Sort by:   newest | oldest | most voted
Pirates Hurdles
Guest
Pirates Hurdles
1 year 6 months ago

Is the next season IP drop actually injury related or just a function of their team’s not going as deep in the postseason the next year?

Brian Sabean
Guest
Brian Sabean
1 year 6 months ago

#evenyear

Ruben Amaro
Guest
Ruben Amaro
1 year 6 months ago

#MaybeNextYear?#MaybeNextDecade?#MaybeNextCentury?

Jason B
Guest
Jason B
1 year 6 months ago

#Maybewhenyouleave

ElJimador
Guest
ElJimador
1 year 6 months ago

Jeff, is there anything in these results to make you think the drop off in the next season had anything to do with the workload of the previous season specifically instead of just normal regression and the aging curve? Seems to me if you set up almost any other parameter besides number of pitches to arrive at a control group of 40 aces having better than average seasons even by their own standards, and then comparing what they did those seasons vs. the next, you’d likely get the same results.

Pirates Hurdles
Guest
Pirates Hurdles
1 year 6 months ago

Isn’t the conclusion essentially that there is nothing to worry about. I don’t think its worth the work to prove it even further.

Cidron
Member
Cidron
1 year 6 months ago

Good question. If it was ‘just the aging’, that would be easily found in any pitcher, whether or not he met the criteria above. Just look at any pitcher of that age group and see his relative regression. If these ones are in line with that group (vs a single pitcher, to increase sample size/data field), then its aging. If its a more dramatic regression, then its whats being addressed here. Pitches thrown in a season.

Avattoir
Guest
Avattoir
1 year 6 months ago

Exactly.

Hey, ace, how’d 2013 go?
– Best ever.
How ’bout 2014? Best again?
– Nope.
Could be a lesson there, ace!
– Like, when one year is best ever, all others ain’t?

stevenam
Guest
stevenam
1 year 6 months ago

Exactly. Thank you.

Emcee Peepants
Guest
Emcee Peepants
1 year 6 months ago

*except if you’re Randy Johnson.

Joe
Guest
Joe
1 year 6 months ago

Whenever you say trust the projections, you are supposed to hyperlink to Dave’s article telling us we should trust the projections.

MustBunique
Member
Member
1 year 6 months ago

Good stuff Zim. I would have intuitively expected a drop off. I had never bothered to check the numbers, though. You have successfully made me reconsider my thought process.

One question, did you consider constraining your range? I would think that 1995 would still be in the heart of what most consider the steroid era. To whatever extent PEDs were used, I would think they would help with pitcher recovery, so the next year after 4000 pitches in 95 might not look the same as in 2011 (randomly chosen). Thanks again.

here goes nothing
Guest
here goes nothing
1 year 6 months ago

I’m not sure I really agree with the conclusion here. I’d think of the true talent level being unchanged–it’s a statistical truism that you’re more than likely going to regress toward a mean, and a 4k pitch season is going to be an outlier in terms of everything other than true talent level breaking well for you.

If Marcel projections represent something like true talent level at the beginning of a season, and year-end results are like Marcel filtered through the chaos of a season…they still beat their Marcel projections, which to me suggests they didn’t do “worse.” I think this is just a question of us thinking of this differently, but I could also just need more coffee.

Kevin
Guest
Kevin
1 year 6 months ago

Chris Carpenter was never the same pitcher again after his 4K pitch season. But he had a history of injuries.

Avattoir
Guest
Avattoir
1 year 6 months ago

‘never the same again’ could just mean different; or could simply come down to, one year will be “best” and all others will not.

ElJimador
Guest
ElJimador
1 year 6 months ago

Carpenter was also 36 that year. Pretty good chance he was never going to be the same again whether he threw 4000 pitches or 4.

ltbailey
Member
ltbailey
1 year 6 months ago

While it would exaggerate the small sample size problem, I’d like to see monthly or first half/ second half splits for the subsequent season to identify some trade opportunities. Sure the projections are close on the season as a whole, but any insight into whether guys tend to start the season poorly and improve over time or vice versa?

CircleChange11
Guest
CircleChange11
1 year 6 months ago

It seems to me that arm injuries follow a season where slider usage jumps up , not necessarily innings pitched or pitches thrown.

it also seems to be a trade pitchers are willing to make due to the effectiveness of the slider.

cs3
Member
cs3
1 year 6 months ago

Just realized how utterly stupid pitcher WAR is.

Lets look at Bumgarner in 2011:
204 IP
191 K’s
22.6% K
5.5% BB%
3.21 ERA
3.17 SIERA
4.6 WAR

and now 2014 Bumgarner:
217 IP
219 K’s
25.1% K
4.9% BB
2.98 ERA
3.05 SIERA
3.6 WAR

so he was better across the board in 2014, but somehow he was worth a full win more in 2011?
To make it clear how ridiculous that is, in 2011 he had ~28% more WAR than in 2014.
Doesnt add up.

a eskpert
Guest
a eskpert
1 year 6 months ago

Dongers and context.

cs3
Member
cs3
1 year 6 months ago

I realize the HR rate is the main culprit in this case that causes the difference. But the fact is that he was significantly better in 2014, regardless of HRs allowed.

WAR for pitchers needs a serious revamp in my opinion. It really tells us nothing. At least not close to as much as hitter WAR

pft
Guest
pft
1 year 6 months ago

rWAR has him 1.6 WAR better in 2014.

fWAR uses FIP. FIP is fine for projections, but WAR should be based on actual results

Nivra
Guest
Nivra
1 year 6 months ago

Don’t forget that 2014 Bumgarner also racked up 1.2 fWAR in batting value.

genghiskhanull
Guest
genghiskhanull
1 year 6 months ago

I know that using FIP in the calculations instead of ERA is a major reason for this, but the run scoring environment has changed significantly. Bumgarner was 21st in the MLB in ERA this season while posting a 2.98. He would’ve been 17th in 2011; go back three more years and 2.98 is in the top ten.

I think the best solution to fix pitcher WAR is to find some sort of compromise between what should have happened and what actually did. Neither FIP nor ERA is entirely reliable to gauge a pitchers value, but I’m pretty sure the truth lies somewhere in the middle.

cs3
Member
cs3
1 year 6 months ago

And you cannot tell me that replacement level was over 28% lower in 2011 tha it was in 2014, because it wasnt.

obsessivegiantscompulsive
Guest
1 year 6 months ago

Love the article, but I’m lost. Jeff said that he was going to use the median value going forward, but then all the values then were the average value instead. Nobody noticed that? Plus, I see 21 less IP, not 23.

And I know I’m going into SSS space, but most of the pitchers in the dataset were on the older side. If limited to just the seasons, say, under 30 (in their 20’s), it would be interesting to see what the numbers look like.

KK-Swizzle
Guest
KK-Swizzle
1 year 6 months ago

According to my (admittedly unscientific) analysis, the median projection for the time at which an “over-used, disparaging, pose-as-Ruben-Amaro” comment appears on fangraphs is approaching 3 (comments that is).

bjoak
Member
bjoak
1 year 6 months ago

It’s a good first step. My concerns about Bumgarner, however, are as follows: He threw 175 pitches in the space of four days (after throwing all the other pitches) and he’s only 25. Do we have anything like a sample to even begin to understand how likely that is to result in an injury?

wpDiscuz