Strange trends in pitches per plate appearance

Here’s a plot showing the average pitches per plate appearance across all of MLB going back to 1988:

Click through for discussion.

Why 1988? That’s as far back as Baseball-Reference.com has complete pitch data.

The graph above shows the average number of pitches (the circles) as well as the average for the team with highest and lowest values for each season (the undotted lines).

What does this mean? Why does this happen? I don’t think  have definitive answers, but here is some food for thought:

  • There is a strong correlation between the number of pitches per plate appearance and the rate of batters striking out. The more pitches a batter sees, the more often he gets to a 2-strike count, and the more often he ultimately strikes out. Think of it this way–if a batter swung on both the first and second pitches of every at-bat, he’d rarely strike out because he’d put the ball in play a lot.
  • It’s possible that this is a sort of Moneyball effect, with patience in working the count becoming more prevalent across baseball. I tend to doubt this, though, as walks haven’t changed much over the last 20-some years.
  • Systematic changes in umpiring could be involved, although I don’t see how.

The one thing that does make sense to me is that pitches per plate appearance correlates with both the total innings pitched and total appearances by relievers. We know that from the late 1980s until the late 2010s, duration of starts got shorter and shorter, meaning that relievers pitched more, on average, in each game. We also know that specialists such as the LOOGY have become more and more prevalent. These trends have started to reverse a bit in the last few years.

I think it’s likely that starting pitchers are much more conscious of pitch counts than relievers. A starter knows that he won’t have the gas to pitch the 7th inning if he’s reached 100 pitches through 5 innings, and this may make him more likely to throw “out” pitches, such as those trying to induce a swing and a weak grounder, as opposed to trying to really work a batter and ultimately strike him out.

Relievers, on the other hand, have no such limitations. Again think of the LOOGY in the extreme case–this is a Lefty One Out GuY who knows he’s only going to face one batter. After he’s thrown maybe 30 warm-up pitches in the bullpen and a few more on the mound, what difference does it make to him if he throws 2, 3, 5, or 7 pitches to the batter? He might as well be as crafty as possible, and if that means nibbling and throwing lots of pitches, so be it.

This effect is minor, but these days it might be multiplied for 3or 4 relievers who are pitching a total of 3 innings, whereas in 1988 it might have been 2 relievers pitching a total of 2 innings. That’s a lot of batters over the course of a season, so even a small effect can show up as a 5% difference over an entire year.


Comments

Strange trends in pitches per plate appearance — 22 Comments

  1. This isn’t really surprising at all. K rates are higher, and those require, by definition, at least 3 pitches. There are 10 PA/G which result in either a walk or strikeout. That’s an all-time high (actually, 2010 was the high – 10.4 – from what I can tell). That’s definitely going to cause an uptick in pitches.

    Additionally, I’ve read Tango mentioning this a number of times: starter, throughout most of baseball history, have thrown about 100 pitches per game. Sure, the deviation used to be wider longer ago (anywhere from 60-140 pitches, rather than the 90-110 we see now), but the average is about the same. But it makes sense that, if pitchers are still throwing the same number of pitches, and the PAs are longer, you need more relief pitching to pick up the slack. At least that makes sense to me, although I don’t have any actual data (beyond my hearsay evidence) to back it up.

    • Your first paragraph isn’t really cause-effect though. Batters strike out more and that’s why there are more pitches per plate appearance? They both seem to be effects, to me–what is the CAUSE?

      • I guess I didn’t really mean it in a cause-effect way. I meant it more in the way that, based on the stats and trends of which I was aware, this piece of information makes a lot of sense. As to an underlying cause? I have no idea.

  2. I’d be interested in seeing the trends in number of pitches/PA by inning over the last 25 years. I bet we’d see a pretty significant increase in innnings 7-9 over that period.

  3. And the gap between the highest and lowest team each year has tightened since 15+ years ago. This could indeed reflect the spread of the Moneyball philosophy of value in seeing alot of pitches and working the count.

  4. Would be interesting also to see correlation to game time duration. I except game time will have increased by more than 5% because of the effect of more pitchers (mound visits, warmups), as well as more pitches.

  5. Another factor suggested by the graph but not mentioned yet is HR rates.

    When HR rates are higher, pitchers tend to be more careful.

    The long upward trend of the main graph begins in 1993-94, shortly after the start of what we now call the PED Era, and ends in 2009, shortly after that era’s end. (Or perhaps I should say, “end.”)

    • That sounds right to me, John. Historical hypothesis: Over baseball history, gloves got bigger, playing fields became better cared for and thus easier to field on and fielders became faster afoot. This made getting on base via balls in play tougher. Hitters adjusted by swinging more for the fences, including adopting lighter, whippier bats that are harder to control with respect to making contact but result in more homers when contact is made. Essentially, batters accepted more swinging strikes (i.e., less frequent contact) in return for more homers, an adjusted tradeoff that makes sense if fielding errors decline and the quality of defense makes finding holes for base hits harder. Weight training and PEDs, which brought the ability to hit more homers more in reach for more players, presumably added to more and more players accepting this tradeoff, as perhaps did trends in stadium architecture. A newer trend toward larger playing fields, plus drug-testing, may at long last be helping to reverse this long-term historical course.

      Note that this hypothesis also helps expalin the long-term trend toward shorter appearances by starting pitchers — more pitches (more swinging strikes, fewer balls in play) means starters tire sooner in the game and the role of relievers grows.

      • birtelcom,

        With scoring going down significantly the last several years, might this mean that starters go somewhat DEEPER into games, and relievers are used a little less frequently? Is the trend towards more relievers used per game inexorable enough, to overcome lower offensive levels?

        • I think the answer to both questions is yes.

          Starters may go slightly deeper, but managers may keep subdividing the remainder among more and more relievers.

          Average start in 2000 was 5.92 IP.
          Average start in 2011 was 6.03 IP.

          Average relief appearance in 2000 was 1.17 IP.
          Average relief appearance in 2011 was 1.02 IP.

          In 2000, there were 140 pitchers with at least 40 games in relief.
          In 2011, there were 163.

        • My guess is that if the scoring decline is accompanied by a decline in pitches thrown and a decline in the length of games then we probably will see the beginning of a trend toward longer outings from starters. A good starter on his 95th pitch (or his 120th minute into the game) may be a better choice than a team’s middle relievers, whereas that same starter on his 105th pitch (or his 135th minute into the game) may not be. Note that declines in scoring are not necessarily accompanied by a decline in pitch count or game times, if the main driver in the scoring decline is yet more strikeouts, or if MLB takes up any time saved through lower scoring games by adding yet more commercial time between innings (g-d forbid!).

        • Average per game for all MLB:
          Number of Relievers Used, Outs per Appearance, and Total Relief Outs per Game:
          2011 — 2.86 … 3.07 … 8.79
          2000 — 2.54 … 3.52 … 8.95
          1990 — 2.02 … 4.26 … 8.58
          1980 — 1.56 … 5.11 … 7.99
          1970 — 1.66 … 4.64 … 7.72

    • kds — I can’t find a way to get that data straight from B-R, but with a little Excel work I figured that relievers in 2011 averaged 3.89 Pit/PA.

      And since the overall average was 3.81, and SPs accounted for about 2/3 of all IP, algebra says that SPs averaged about 3.77 Pit/PA.

      (Since I was using season data for individual pitchers, I set the threshold for being counted as “reliever” at relieving in 80% of his total games.)

      I’m not sure if that gives you enough for what you wanted … I kind of lost sight of the point, and now it’s late and my brain’s getting fuzzy.

    • Fangraphs has the data.

      For 2011:

      Starters threw 470896 pitches while facing 124544 batters. That’s 3.78 pitches per plate appearance.

      Relievers threw 236948 pitches while facing 60701 batters. That’s 3.90 pitches per plate appearance.

      I guess me theory holds water.

      If you want to check earlier seasons, here’s where to find the data on Fangraphs:

      From the main page, click on “Leaders”
      Then click on “League Stats”
      then “pitching”
      Then click on “starters” or “relievers” to get the data for each group.
      Pitches appears under the “batted ball” tab, and total batters faced appears under “standard”

  6. It is a shame that we don’t have ball-by-ball for the early days. I suspect that there are all sorts of subtle changes going on here, coming together to create this effect.

    I had the thought that this change may be similar to the dead-ball/live-ball change. It was an era dominated in the first half by pitchers, then by bats. Looking at one extra-ordinary pitcher who crossed the boundary, one W. Johnson. It was interesting to look at his BF/IP and completion rates (starts that became complete games) over his career:

    Dead Live
    BF/IP 3.85 4.21
    Comp Rate 0.884 0.633

    This being split at 1920. Obviously this doesn’t mean too much in terms of this argument, but I suspect that the trend that he shows — more batters per innings and lower completion — would be seen across the board. This amounts to more pitchers per batter and therefore shorter outings.

    I suspect that the trend that Andy shows is a cyclic one with the trough in the late 60s early 70s.

  7. There may not be one single cause, but multiple factors. More offense/HRs causes pitchers to be careful (noted by John A above), changing their pitching patterns. Perhaps a slight narrowing of the strikezone by umpires.

    Is there a difference between leagues? Wondering if the DH impacts the average.

  8. I believe this has more to do with the Moneyball effect than you might think, but based on the changing perceived cost of strikeouts instead of the changing perceived value of walks. One of the earlier Bill James learnings was that a strikeout isn’t a worse out than other outs. No, you don’t advance runners, but you also avoid double plays. As the stigma of striking out has evaporated, batters are not swinging at borderline pitches to protect a possible called strike because they are not afraid of striking out. I would be interested in seeing swing rates by pitch count to see if batters are swinging less than they used to, and if the rate of called strikes per plate appearance has gone up.

  9. I don’t get the part about “if a batter swung on both the first and second pitches of every at-bat, he’d rarely strike out.” If nothing else, the fact that there *is* a second pitch means he didn’t get a hit on the first pitch, meaning he *always* goes into the second pitch down 0-1. Then, if he should happen to fail to hit the *second* pitch for a hit, a foul, an error, or an out, necessitating a *third* pitch, he *always* goes into the third pitch 0-2. I don’t see how always going into the third pitch of an AB 0-2 is going to lower his strikeouts.

Leave a Reply

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