What’s In a Young Pitcher’s Strikeout Decline?

Strikeouts for pitchers aren’t overrated, and here’s why: there is no more fundamental indicator that a pitcher is or isn’t hard to hit. To get strikeouts, you have to throw strikes or pitches that look like strikes, and the batter has to not put those pitches in play. All that’s important for a pitcher to do goes into the generation of strikeouts. There are, of course, relatively ineffective pitchers who get strikeouts. There are relatively effective pitchers who do not get so many strikeouts. Strikeouts aren’t everything, because the name of the game isn’t “Get The Most Strikeouts”, but they are the closest to everything of any of the basic stats. Make people miss and you’re probably good.

So if you want to find pitchers people are buzzing about, follow the strikeouts. If you want to find young potential aces, follow the strikeouts. Yovani Gallardo seemed like a young potential ace. He’s still not old, and he’s still plenty talented. But people have been waiting for him for years. He has yet to post a full-season ERA- under 90. His runs have never quite matched his peripherals. And this past season was something of a worrisome mess. In 2009, Gallardo had baseball’s seventh-highest strikeout rate, essentially equal with Clayton Kershaw. He kept on striking out about a quarter of batters through 2012. This last year, his strikeouts matched Jordan Zimmermann and Edinson Volquez. His rate was still fine, but considerably worse, and it’s enough to make one wonder: what happens when a young starter loses strikeouts?

It would be easier to explain with older starters. They could be hurt, or they could just be reaching the end of the line. Every career dies somewhere, and strikeouts are usually one of the first things to go. We don’t generally expect people to start declining in their 20s, so when you have something like a strikeout decline for a younger starter, is it a blip, or is it an indicator? That is, do pitchers bounce back, or do they just find their new levels?

Gallardo, as it happens, isn’t the only young starter to be coming off a strikeout decline, where “young” is defined as “under 30”. Between 2012-2013, Gallardo’s strikeout rate lost 5.1 percentage points. Stephen Strasburg, David Price, and Dillon Gee each lost 4.1 percentage points. But Strasburg had a big second half, and might’ve been cutting his strikeouts deliberately to get quicker grounders. Price claims to have been working for efficiency, and he countered the strikeout decline by throwing strikes almost exclusively. Gee actually got fewer strikeouts in the second half, so he’s a guy to keep in mind here, too. But Gee already had his shoulder operated on in 2012, and he fought a known elbow issue as well. Gallardo’s never missed time with an arm issue, yet last year his velocity was down across the board. With Gallardo, there was the same workload, but less stuff and fewer whiffs. There’s no apparent easy explanation. In this way Gallardo stands alone.

Let’s go over some history. The first thing we’ll do in the leaderboards is select for starting pitchers. Then we’ll set an innings minimum of 100, and we’ll make sure to only look at guys on the bright side of 30. I took the leaderboards back to 2000, and then I looked for pretty big year-to-year strikeout declines. I somewhat arbitrarily set my dividing line at -4 percentage points. I was left with a pool of 63 pitchers. To repeat all that: there were 63 pitchers

  • as starters
  • since 2000
  • who threw at least 100 innings in consecutive years
  • while younger than 30
  • and had their strikeout rates decline at least four percentage points

Then it was simply a matter of seeing how those pitchers did the next year, the year after losing so many strikeouts. Here’s a table of names and some of the information:

You Aren't a FanGraphs Member
It looks like you aren't yet a FanGraphs Member (or aren't logged in). We aren't mad, just disappointed. We get it. You want to read this article. But before we let you get back to it, we'd like to point out a few of the good reasons why you should become a Member. 1. Ad Free viewing! We won't bug you with this ad, or any other. 2. Unlimited articles! Non-Members only get to read 10 free articles a month. Members never get cut off. 3. Dark mode and Classic mode! 4. Custom player page dashboards! Choose the player cards you want, in the order you want them. 5. One-click data exports! Export our projections and leaderboards for your personal projects. 6. Remove the photos on the home page! (Honestly, this doesn't sound so great to us, but some people wanted it, and we like to give our Members what they want.) 7. Even more Steamer projections! We have handedness, percentile, and context neutral projections available for Members only. 8. Get FanGraphs Walk-Off, a customized year end review! Find out exactly how you used FanGraphs this year, and how that compares to other Members. Don't be a victim of FOMO. 9. A weekly mailbag column, exclusively for Members. 10. Help support FanGraphs and our entire staff! Our Members provide us with critical resources to improve the site and deliver new features! We hope you'll consider a Membership today, for yourself or as a gift! And we realize this has been an awfully long sales pitch, so we've also removed all the other ads in this article. We didn't want to overdo it. In case you didn't read all that, here's the short version. Membership includes:
✓Ad-free browsing
✓Unlimited FanGraphs & RotoGraphs articles
✓Historical & platoon projections
✓Leaderboard & player-page stat heatmaps
✓One-click data exports
✓Customizable player-page dashboards
✓Dark & Classic site modes
✓Weekly Members-only mailbag
✓FanGraphs Walk-Off, your year in review
✓Optional homepage photo removal
✓Gift articles to non-Members
✓Full postseason coverage
Pitcher Year GS FIP- K% Year X+1 GS FIP- K% Year X+2 GS FIP- K%
2000Bartolo Colon 30 80 26% 34 90 21% 33 86 15%
2001Bartolo Colon 34 90 21% 33 86 15% 34 90 18%
2001CC Sabathia 33 95 22% 33 89 17% 30 90 17%
2001Chan Ho Park 35 93 22% 25 108 18% 7 153 11%
2001Glendon Rusch 33 90 20% 34 110 15% 19 91 16%
2001Javier Vazquez 32 70 23% 34 85 18% 34 74 26%
2001Julian Tavarez 28 97 15% 27 114 9% 0 78 11%
2001Kerry Wood 28 81 29% 33 94 24% 32 85 30%
2001Mike Hampton 32 99 14% 30 112 9% 31 94 13%
2001Paul Wilson 24 104 17% 30 115 13% 28 109 13%
2001Roy Halladay 16 50 23% 34 66 17% 36 70 19%
2001Shawn Chacon 27 101 19% 21 127 13% 23 87 16%
2001Terry Adams 22 75 20% 19 96 15% 0 66 18%
2002Barry Zito 35 91 19% 35 91 15% 34 100 18%
2002Damian Moss 29 115 15% 29 139 10% 2 128 17%
2002Matt Clement 32 81 25% 32 95 20% 30 91 25%
2003Brandon Webb 28 73 23% 35 95 18% 33 78 18%
2003Dontrelle Willis 27 83 21% 32 95 16% 34 72 18%
2003Javier Vazquez 34 74 26% 32 108 18% 33 91 21%
2003Kerry Wood 32 85 30% 22 85 24% 10 113 26%
2003Randy Wolf 33 101 21% 23 101 15% 13 114 18%
2004Ben Sheets 34 59 28% 22 77 22% 17 53 27%
2004Bronson Arroyo 29 83 18% 32 99 11% 35 90 19%
2004Eric Milton 34 119 19% 34 126 14% 26 116 14%
2004Freddy Garcia 31 82 21% 33 91 16% 33 98 15%
2004Gil Meche 23 112 18% 26 123 13% 32 106 19%
2004Jason Marquis 32 106 16% 32 118 12% 33 134 11%
2004Joel Pineiro 21 102 19% 30 107 13% 25 124 10%
2004Oliver Perez 30 77 30% 20 149 21% 22 127 19%
2004Victor Zambrano 25 111 19% 27 105 14% 5 144 16%
2005Josh Beckett 29 79 23% 33 109 18% 30 68 24%
2005Mark Buehrle 33 77 15% 32 112 11% 30 93 14%
2005Noah Lowry 33 91 20% 27 109 12% 26 109 13%
2005Roy Halladay 19 69 20% 32 79 15% 31 81 15%
2006Ben Sheets 17 53 27% 24 93 18% 31 79 20%
2006Daniel Cabrera 26 92 24% 34 112 18% 30 127 12%
2006Jered Weaver 19 87 21% 28 92 17% 30 91 20%
2006Scott Olsen 31 98 22% 33 119 16% 33 115 13%
2006Vicente Padilla 33 91 18% 23 115 13% 29 113 17%
2007Daniel Cabrera 34 112 18% 30 127 12% 9 143 8%
2007Johan Santana 33 88 27% 34 83 21% 25 92 21%
2007Roberto Hernandez 32 91 16% 22 115 11% 24 123 13%
2008Ervin Santana 32 77 24% 23 117 17% 33 107 18%
2008John Lannan 31 111 15% 33 111 10% 25 112 11%
2008Micah Owings 18 104 19% 19 136 12% 0 114 23%
2008Scott Kazmir 27 104 26% 26 101 18% 28 145 14%
2009Brett Anderson 30 87 20% 19 79 16% 13 101 17%
2009Jake Peavy 16 75 27% 17 92 21% 18 80 19%
2009Matt Garza 32 99 22% 32 111 18% 31 74 24%
2009Scott Kazmir 26 101 18% 28 145 14% 1 462 0%
2009Zack Greinke 33 53 27% 33 79 20% 28 76 28%
2010Francisco Liriano 31 64 25% 24 113 19% 28 104 24%
2010Jason Hammel 30 82 18% 27 109 12% 20 77 23%
2010Jered Weaver 34 76 26% 33 83 21% 30 94 19%
2010Jhoulys Chacin 21 77 23% 31 96 18% 14 114 14%
2010Ricky Nolasco 26 95 22% 33 90 17% 31 99 15%
2010Travis Wood 17 84 21% 18 106 16% 26 119 18%
2011Brandon Morrow 30 88 26% 21 86 21% 10 134 17%
2011Chris Volstad 29 111 16% 21 126 12% 0 115 6%
2011Ricky Romero 32 102 19% 32 122 15% 2 206 16%
2011Tommy Hanson 22 97 26% 31 118 21% 13 126 17%
2011Ubaldo Jimenez 32 87 22% 31 127 18% 32 90 25%
2011Zack Greinke 28 76 28% 34 78 23% 28 90 21%

In Year X+2, four pitchers didn’t start a single game in the majors. Six more wound up throwing fewer than 50 innings as starters. Of the remaining 53, 19 had further strikeout-rate declines, while 34 bounced back. Nine exceeded their strikeout rates in Year X.

Let’s play around with those 53, understanding that by doing so we’re already eliminating ten guys from the sample who didn’t do so hot. Some average performance numbers:

YEAR X

  • 29 starts
  • 180 innings
  • 86 ERA-
  • 86 FIP-
  • 22% strikeouts
  • 8% walks

YEAR X+1

  • 29 starts
  • 181 innings
  • 102 ERA-
  • 101 FIP-
  • 17% strikeouts
  • 8% walks

YEAR X+2

  • 27 starts
  • 169 innings
  • 100 ERA-
  • 98 FIP-
  • 18% strikeouts
  • 8% walks

As a group, plenty of these pitchers continued to pitch often, but they didn’t bounce back very much to old levels. After losing about five percentage points of strikeout rate, they gained back just one. So they wound up with sustained losses in both ERA- and FIP-. They went from being good starters to average starters, overall, and there are as many nightmares as success stories.

Credit to Ubaldo Jimenez for bouncing back in 2013. Jason Hammel rebounded in 2012, although he was also newly out of Colorado. Zack Greinke’s strikeout rate shot up between 2010 and 2011. But then, Greinke also lost a ton of ground between 2011 and 2012, and he lost only more ground between 2012 and 2013. Scott Kazmir’s strikeouts plummeted before he disappeared. Daniel Cabrera turned into a catastrophe. Tommy Hanson’s slippage hinted at doom. Brandon Morrow is a total mystery. It’s not the most encouraging group of pitchers, if you’re looking for Gallardo to get back to what he was.

What’s positive is that Gallardo has a long, established track record of getting strikeouts. He’s still not old, he made just about every start a season ago, his stuff still looks more or less like his stuff, and he didn’t have any unusual problems with control. But, he did lose velocity. He did lose strikeouts, with the corresponding increase in contact allowed. Recent history shows that, when a young starting pitcher loses a bunch of strikeouts, more often than not they don’t really come back. Gallardo might have a physical problem, or he might just be declining. But it seems like the probability is that he is what he most recently was. If it’s any consolation, that means Gallardo might no longer be a frustrating ought-to-be ace. That he might just be differently frustrating is whatever the opposite of consolation is.





Jeff made Lookout Landing a thing, but he does not still write there about the Mariners. He does write here, sometimes about the Mariners, but usually not.

24 Comments
Oldest
Newest Most Voted
pft
12 years ago

What should we make of Tanakas declining K rate in the NPB?

pft
12 years ago
Reply to  Jeff Sullivan

Their K rate was in line with their career averages, and they were in their 30’s when they came over.

Iwakuma k rate did drop off at age 25 but he was injured that same year. His K rate was stable when he went to the Mariners in his 30’s. Kuroda’s K rate increased at 25 before dropping off at age 28 and he was injured the next year, but bounced back the year after. Kurodas k/rate was declining when he came over but that was gradual and looks like age related decline.

Maybe its nothing but I always feel a bit nervous when a 25 yo’s K rate declines.

pft
12 years ago
Reply to  pft

Actually, Tanaka was only 24 last year and 1.8 k/9 off his peak at age 22

AK7007
12 years ago
Reply to  Jeff Sullivan

I’ve also heard anecdotes that there is a philosophical difference between here and Japan – pitchers are taught to basically only throw strikes (a la 2013 David Price), which makes predicting their K-rate here not an exact science. Is there data on Zone% available for NPB?

DaveC
12 years ago

Something interesting here is also that Gallardo saw his GB% increase and his HR rate decrease. Perhaps he’s throwing more sinkers and/or lower in the zone and essentially sacrificing strikeouts for grounders.

Andrew J
12 years ago

Seems like this is the definition of regression to the mean. Player strikes out abnormally amount of people one year, and is very good, then goes back to being average, because it is just hard to keep up above average performance.

Well-Beered Englishman
12 years ago
Reply to  Andrew J

While this is in one sense very true, we should caution that if players regress to a mean, it is to their generalised talent level, not to league averages. A player performing significantly better or worse than league average does not suggest a future change in performance, by itself.

MGL
12 years ago

“A player performing significantly better or worse than league average does not suggest a future change in performance.”

Yes, of course it does. Knowing nothing else about a player other than a certain rate stat, such as K%, if he is above league average, he is expected to decline and if he is above league average, he is expected to improve, assuming he gets to play the second go around, and not including aging decline or improvement.

Try it for any stat for any large group of players (to avoid noise). You will see that it works nearly 100% of the time!

That is regression toward the mean. Regressing “towards someone’s talent level” is misunderstood. In fact, there really is no such thing as “regressing toward (or to) your own talent level.”

A player does not regress toward his own talent level. He is expected to perform at EXACTLY his talent level. Unfortunately we never know exactly what that talent level is. So we estimate it by using his observed performance regressed toward league average performance, where “league average” is the average of all players in the population that we think the player comes from. That could be “the league” or it could be “all RH starting pitchers.” Or it could be “All LH starters who throw 93 mph.” Etc.

siddf
12 years ago
Reply to  MGL

this is not correct

Tim
12 years ago
Reply to  MGL

If you really believe this then you have to address the question of what is so magical about MLB league average, which is not at all “the mean.” If we expect players to regress to some sort of global mean, then every MLB player would be expected to get worse, and 34-year-old town ball guys would occasionally hit like Mike Trout.

“A player does not regress toward his own talent level. He is expected to perform at EXACTLY his talent level.”

This is not at all true. His talent level is expected to be the centerline for his variance. A player does not, in fact, regress towards anything. (Except in the sense that the average historical baseball player is deceased.) A player’s performance will regress to his talent level if his talent level remains unchanged.

mch38
12 years ago
Reply to  MGL

Don’t make things up; let me google that for you.

http://lmgtfy.com/?q=regression+toward+the+mean

Brad JohnsonMember
12 years ago
Reply to  MGL

It’s worth pointing out that most of the stats we use are imperfect at measuring “performance” where performance is defined as how well a player actually played. Even a stat as tightly composed as wOBA includes components that aren’t really performance. This makes some of the statistical analysis we do a little tricksy.

As an example, if I go 0-for-4 you’d say I performed poorly. So would wOBA. If however, I scalded 4 line drives at the shortstop, then I actually performed well with bad results. Over a large sample this mostly gets balanced out and we have a few techniques to further adjust for this “noise,” but ultimately it still affects the data and what we should expect.

August FagerstromMember since 2018
12 years ago
Reply to  Andrew J

Except the study wasn’t just on high-strikeout pitchers. There were pitchers with already low strikeout totals that still declined in the study.

BMarkham
12 years ago

Very interesting article. I think it’s important to note (and you touched on it a bit) though that Price is different then the pitchers in your study. Price’s strikeout rate did drop in 2013, but he also had his best BB/9 last year as well his best K/BB. Price seems to have made a conscious decision to throw more strikes and take less strikeouts in exchange for much fewer walks and weaker contract. The pitchers in your study averaged a BB% of 8 in all three years so it doesn’t seem like they really pursued the same strategy.

It seems like a strategy pitchers just have to implement once they lose speed, whether it be age-related decline or what else. Those pitches just out of the zone are going to fool hitters Y% less of the time when you lose X% of velocity. Less of a margin of error. Command and the edge of the plate is the name of the game.

pft
12 years ago
Reply to  BMarkham

Price changing to a pitch to contact hitter is what Pineda says he did in the 2nd half of 2011 to explain his drop off, and then he needed labrum surgery after an awful ST in 2012. When pitchers who have succeeded as power pitchers change to pitch to contact, its usually a sign something is wrong. That Price changed his approach after a bad start and 6 weeks on the DL is troubling, even if he pitched well in the 2nd half. Some of the success was catching teams by surprise with his new approach.

BMarkham
12 years ago
Reply to  pft

Meh, he posted a better FIP in 2013 than every year except his Cy Young year, even with the slow start. Doesn’t sound troubling to me. In addition to Wainwright and Lee (two guys way up there in K/BB by the way)Greinke is another guy who has traded a few Ks for much less walks and pitches thrown. It’s just good strategy, especially when you’re reaching that point in your career where you may see a velocity drop. The best past their prime pitchers have always been guys with exceptional command (Maddux, Glavine, Moyer, Colon). Price just seems to be getting ready for it a little ahead of the curve.

AddyMac
12 years ago
Reply to  BMarkham

There probably is a way to combine both points made by “MGL”…and of course, my boy “pffft”…

Making an estimation of future performance or decline based off one rate stat allows for a lot of other variables (separating the 53 pitchers left into various other groups–IE, handedness, stuff, previous velocity, previous ranges of FIP-…IN ADDITION TO declining-K-rate-by-4+%)…

And, in addition to that point, when it comes to separating out guys like Price (who see a decline in their K rate because of a change of philosophy)–only using one rate stat such as strikeout percentage also takes away the ability to filter the “types” of K-declining pitchers even more.

To gauge whether or not the sample size of 53 could be shaved down even more to eliminate easier to explain young, declining-in-strikeout types fitting into the less-strikeout-more-IP-and-groundball/quick outs group, I’d imagine looking at things like total strikes thrown%, pitches/batters faced for the whole year, strike% with fastball, etc could be used.

Given that this is a daily article (a very good one), though, that would be an interesting–albeit larger–data project. Really cool stuff here. I also wonder if total IP in career before the “year X” where the pitcher initially dropped in strikeout percentage also could create another “group” or “type” of this pitcher younger than 30 worth examining, of the final sample size of 53.

CircleChange11
12 years ago
Reply to  BMarkham

Cliff Lee did the same thing didnt he?

Basically decided to stop walking guys.

Adam Wainwright in 2013?

Colin
12 years ago
Reply to  CircleChange11

Did he ‘decide’ to do that or just get better at it?

Jeff Long
12 years ago
Reply to  Colin

It’s a clear decision in my mind. Take a look at a pitcher’s approach. For example, Cliff Lee pounds the bottom of the strikezone with pitches that are more or less easy to command (sinkers, etc.). Other pitchers don’t take the same approach, and see increased walk rates as a result.

This is true for both Lee:
http://www.beyondtheboxscore.com/2013/10/11/4826428/the-art-of-not-walking-batters

and Wainwright:
http://www.beyondtheboxscore.com/2013/9/13/4725488/a-pitchf-x-profile-of-adam-wainwright

Frank
12 years ago

Is there something to the recent trending up of the changeups popularity. Its not like the quality of Prices pitches sufferd overly much over his early years
. His ground ball rates up and strikeouts down but the increase in balls in play is a result of a more groundball oriented approach. If he didnt lose his curve this season he would have been deadly. The list doesnt really show a legitimate sample of left handed pitchers, in an already small sample size.

MGL
12 years ago

There are many issues unmentioned here. Lots of these pitchers pitched well above the league average for starters, so for most of them it was inevitable that they would “lose” some of that K rate in any given year. What would we expect the next year? Simply their career numbers regressed toward some mean, around 15-16% (that is league average).

Some of these pitchers had many years of a high K rate, so that their year after the decline, we expect a large bounce back. Some of them had only a year or two of a high K rate followed by a lower K rate. We expect these pitchers to bounce back but not as much because they would be regressed toward the mean more.

Some of these pitchers lost K rate because their velocity declined. Some lost it for other reasons (luck, different style of pitching, different repetoire, less movement, etc.).

To use all these pitchers to try and figure out what Gallardo is likely to do is a waste of time. It is a one-size fits all solution to a problem which has lots of alternative and much better solutions.

For example, with Gallardo, we have many years of a K% in the low to mid 20’s. That suggests that he will tend to bounce back into the low 20’s. That would be completely different from a pitcher who, say, has one year of 25%, then one year of 17%.

We also have a pitcher who is 28 years old. We typically see velocity and K rate decline in 28 year old pitchers.

And most importantly, we see a decline of around 1 mph 2 years in a row. Unless, almost miraculously, he gains back 2 mph on his fastball, we can expect that his days of striking out 25% of his opponents are gone.

So if you want to create a list of “comps” at the very least use starters in their late 20’s with many years of a sustained high K rate, and who lost a significant amount of velocity when their K rate declined.

I would venture to say that if you created 2 groups of pitchers – one where their K rate declined at last 5% with no decline in velocity and one where they lost at least 1 mph, you would see a drastic difference in their K rate “bounce back” in year X+2. Someone try it!

AddyMac
12 years ago
Reply to  MGL

MGL–Idk if you saw my comment above. I agree with you. The fact this project can spin off so many other numerous ideas about how to “expand”–or, in an alternative way–“contract”, the sample size group of 53 so different “types” of the 53 pitchers could be grouped together in sample sizes easier to compare, probably means it’s a good start.

Making the groups (of the final 53) based on things like age(+ / – nine years between 20-29 is a wide age range and career IP range..), career GS up to “year X” where the pitcher lost strikeouts, higher strike% and lower pitches-per-AB group (to explain why David Price is on this list), etc etc etc would definitely make this a much more telling and educational study. It’s just a longer statistical data project is all.