Young Pitchers and Complete Games

A couple of weeks ago, I was helping fellow writer, Chris Cwik, look for possible reasons James McDonald fell off the cliff in the second half of the 2012 season. As Cwik pointed out, the 27-year-old McDonald started falling apart after he threw 122 pitches for his only complete game (CG) of the season. I decided to see how young pitchers performed after throwing just one complete game during a season.

I know pitchers, and their managers, would love to go out every night and throw a complete game. With the increased patience batters have in the league now, complete games are becoming rarer and rarer. When a young pitcher has a chance for a complete game, especially a shutout, managers will often give the kid every opportunity for the achievement.

Back 2001, Chad Durbin, who was with the Braves last season, was a promising starting pitching prospect with the Royals. He was 23 years old and in his first full season. Going into the July 28 game against Oakland, he was struggling a bit with a 6-9 record and a 4.76 ERA. On that night, he ended up throwing 137 pitches for a complete game (not even a shutout). After that point, he was never the same. His ERA jumped to 5.59 over the rest of the season and he spent 2002 on the disabled list. He was released by the Royals at the end of 2002.

I don’t like to base conclusions on a sample size of two with one case being 10 years old. To get an idea of a young pitcher’s results after they throw a CG, I selected the following players:

* 27 years and younger
* 1 CG in the season. I limited it to one game to have a nice before and after set of stats.
* The CG wasn’t in the first 2 or last 2 games of the season.
* The CG went 9 innings

Note: Before going on, I looked at two other ideas, did the number of pitches in the CG matter and did their disabled list chances increase. The number of pitches thrown in the CG didn’t matter at all with my sample. Second, not an abnormal amount of the pitchers went on the DL.

From 2011 and 2012, 30 pitchers met this criteria. I looked at their stats before the CG, the first five games after it, and over the rest of the season. I collected the average change and the median change. Because of the small sample size of games before and after the CG, the median value may be more useful. The CG itself was not used in any of the calculations.

Let’s start with ERA.

Time frame: Average ERA, median ERA
Before: 3.91, 4.05
After (1st 5 games): 4.53, 4.55
After (Rest of Season): 4.26, 4.23

Immediately after the CG, the pitcher saw around 1/2 point jump in ERA and 1/4 of a point over the rest of the season. Now, time to look for some causes starting with strikeouts and walks

Stat Change From Before CG: Next 5, ROS
K% (average): +.95%, +.65%
K% (median): -.24%, -.21%

BB% (average): +.17%, +.54%
BB% (median): + .22%, +.27%

By looking at the median values, there is some change for the worse, but nothing close to explain the large increase in ERA. Now, here are the changes in batted ball data:

Stat Change From Before CG: Next 5, ROS
BABIP (average): +.026, +.018
BABIP (median): +.022, -.001

HR/9 (average): +.26 +.29
HR/9 (median): +.07, +.19

Those numbers begin to explain the changes. The pitchers got hit around a little more after the CG. Their BABIP before the CG averaged .288 and it grew to .314 for the first five games and then back down to .305. Their HR/9 went from .86 to 1.14 back to 1.07. It seems like before the CG the pitchers were a little lucky and were a little unlucky after the game.

Putting all the data together, it looks like pitchers going into the CG had been a bit lucky on batted balls. They seemed to be pitching better than they actually were. The manager let them go long for the complete game. For a few games, the pitchers performed worse and then they got back into a semi-normal groove. Using this data, an fantasy owner can try to target pitchers who may seem to be struggling after a CG and pick them up on the cheap.

If a young pitcher throws a complete game during the season expect them to perform worse after the game. Most of the time, they will just regress back to their mean/expected performance. During this time, these players may end up being picked up at a discount.




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.


12 Responses to “Young Pitchers and Complete Games”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. Interesting study. Is there any reason to think a pitcher has an ERA below his true talent in the games before a complete game? Might an alternative explanation be that the complete game impacted his effectiveness in both the short and long term?

    Also, I think these results put a lie to the notion that BABIP is always the result of luck. If BABIP has a strong tendency to rise after a complete game, seems to me this indicates a “true” driver of BABIP, related to arm health.

    Vote -1 Vote +1

    • Jeff Zimmerman says:

      I am with you Dave. I think pitchers are at their worst when they are in a transition stage with velocity, pitch mix, consistency of pitches in the zone, etc. BABIP seems to be one main factors that increases since the pitcher can’t throw to the corners and has to put too many across the middle of the plate.

      The theory goes as follows:
      Stage 1: Cruising along all good and happy at 100% production
      Stage 2: Shoulder hurts when throwing 96 mph fastball. Start to throw slower for less pain or inconsistency and deal with pain. Production during this time is 60%.
      Stage 3: Finally figure he can throw pain free at 94 mph and is now at 80% production.

      It is something I plan on working on in the near future.

      Vote -1 Vote +1

    • Ceetar says:

      Perhaps advanced scouting more than fatigue? If a guy comes to town having just thrown a complete game, the opposition might put a little bit more work into reading up on him and watching tape and all that. He’s found something he’s had some success with, and if he’s getting results, he’s not going to switch it up as much, making it much more exploitable. As teams catch on and he doesn’t have this success he starts making changes that again begin to fool batters. It seems like there might be a cyclical nature to it, and how fast a pitcher adjusts probably plays into how good he is overall.

      Which pitches were these? mostly middling guys or were there aces in there? Maybe it’s not ‘young’ pitchers but ‘average’ pitchers? Curious what the numbers look like for all pitchers. Also curious what happens if you take out the 1CG thing. if the sample around said CG happens to include another one, that should play in.you may get a pitcher twice or more in the same season with overlap, but it seems like you’d be eliminating better pitchers by limiting it to one.

      Vote -1 Vote +1

      • Jeff Zimmerman says:

        Sorry, I should have included the pitchers in the article for reference:

        2012
        Chris Sale
        Ian Kennedy
        Phil Hughes
        Justin Masterson
        James McDonald
        Lucas Harrell
        Tommy Milone
        Kyle Kendrick
        Scott Diamond
        Anthony Bass
        Henderson Alvarez
        Brandon Beachy
        Liam Hendriks
        P.J. Walters
        Neftali Feliz
        2011
        Tim Lincecum
        Ian Kennedy
        Matt Cain
        Justin Masterson
        Chad Billingsley
        Edwin Jackson
        Alexi Ogando
        Jordan Zimmermann
        Vance Worley
        Francisco Liriano
        Wade Davis
        Jo-Jo Reyes
        Carlos Carrasco
        Brad Bergesen
        Zach Stewart

        2011 seemed a little more star loaded than 2012.

        Vote -1 Vote +1

  2. Derek says:

    The HR/9 numbers are the most interesting to me here. Two questions re: that:

    1) Are the pitchers’ flyball rates going up after a CG, too?

    2) Are the pitchers altering (e.g. lowering) their release point after a CG?

    Interesting stuff here. Thanks, Jeff.

    Vote -1 Vote +1

  3. Nivra says:

    In the first 5 games after the CG, you have 30 pitcher-seasons, or 150 total games worth. At 7 IP average, that’s 1050 IP. Is that even enough of a sample to determine a trend?

    What if I randomly took 30 pitchers, and for each pitcher, took a game during the season that’s not the first two or last two, and then did the exact same analysis. What kind of variance would I get in the subsequent statistics?

    Is the ERA increase you see even significant? BABIP increase significant?

    All this could simply be noise supporting anecdotal observations.

    Vote -1 Vote +1

  4. The Humber Games says:

    There could also be a psychological side as well…most complete games involve some degree of batted ball ‘luck’. Maybe young pitchers get away with some bad pitches, get rewarded for it with a complete game, and then find out the hard way that it’s not going to work every time and have to adjust.

    Hard to measure, but I wouldn’t be surprised if you saw some behavior changes from the pitcher (selection, location) immediately after a cg. ‘This was working for me during my cg so I’ll keep doing it’ sort of thing.

    Vote -1 Vote +1

  5. ncb says:

    Just wondering if the complete game is causal. Don’t young pitchers usually end up getting hit harder during the course of the year as the league figures them out and the weather warms up? If you did the same analysis for young pitchers that didn’t throw a complete game during the course of the year (by picking a random game), do you think it would be different?

    Vote -1 Vote +1

  6. rubesandbabes says:

    1) Author states:

    “Putting all the data together, it looks like pitchers going into the CG had been a bit lucky on batted balls. They seemed to be pitching better than they actually were. The manager let them go long for the complete game.”

    Okay, this is saying a couple of things:

    aa) Baseball stats are bigger / more true than the actual game itself. “They seemed to be pitching better than they actually were.” Answer: no, they were playing baseball.

    bb) This can be read as an on-high criticism of the typical methods used by the various MLB managers/coaches to decide when to remove a pitcher from the game. It implies that an MLB manager would just naturally be fooled by what he was seeing because he would never take into account BAPIP (or BAPIP components). Yeah, I think no, not in the era of Strasburg shutdowns.

    2) The two 2012 random guys I looked at Tom Milone and Chris Sale did not support the results of the findings at all. Milone got on a roll after his one CG, and Sale’s CG was in the middle of a good long stretch. (They screwed up the curve with their curves. Haha.)

    3) From the comments: It’s highly questionable to say pitchers build toward complete games in their previous starts. Okay, but fun to look at. It is another thing entirely to suggest that pitchers would pitch not as well in their starts leading up to complete games. Astrology, people – Earth is not in the center.

    4) Glad to see Durbin is still around.

    Sorry criticisms, only coming with good reading at Fangraphs – Happy New Year everyone!

    Vote -1 Vote +1

  7. jj says:

    I’m not clear why you use SP that only have 1 CG, wouldn’t that by default mean you are getting guys who had just one game where they were deemed good enough to go 9IP, and thus skew the results a bit to pitchers who were likely not as good as that one game would indicate?

    Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Current ye@r *