Strikeout pitchers and pitch counts

Another little gem from Bill James online. You know how it's taken as gospel that pitchers need to pitch to contact to lower their pitch counts, and therefore go more innings, right? And how young fireballers will try to strike out fewer batters as they mature to save their arms, right?

Well, it's wrong. High strikeout pitchers, on average, throw only a fraction of a pitch per inning more than low-K pitchers:

Using the top five starters on each team (based on games started), we get 150 Major League Baseball pitchers from 2007. We divide them into three groups of 50 based on their Strikeouts Per Nine ratio. Here is how it comes out:

                                  Innings      BB/9  K/9   Pitches/InningHigh Strikeouts             8,716         3.1     8.1      16.28Medium Strikeouts        7,442         3.0     6.2      16.19Low Strikeouts              7,332         2.8     4.6      15.81

So there is a pattern. Strikeout pitchers do throw more pitches. But it's minimal. Over the course of 7 innings, the high strikeout group throws 114 pitches, medium throws 113 and low throws 111. That's a difference of only three pitches per game from the top group to the bottom. Which, interestingly coincides pretty much with the extra pitches thrown on the 3-4 extra strikeouts those pitchers get.

So, we're back to where we started. Strikeout pitchers do throw extra pitches, but not very many, and not anywhere near as many as most baseball observers might think.

It’s a matter of perspective, I suppose. From now on, when I hear somebody say that strikeout pitchers throw more pitches because the strikeout requires three pitches to get an out, I plan to say, “that’s really not true.” If you want to interpret the data differently, I guess that’s up to you.

High K pitchers tend to be more effective than low-K pitchers, so they retire batters more frequently and they get out of innings just as quickly. Also, walk rate is nearly unconnected to strikeout rate in modern baseball. This was probably a bit different in the past, but today there is almost no correlation between K rate and walk rate.

So, next time you hear somebody saying that the only way Daniel Cabrera will become an innings-eater and a real pitcher is to learn to pitch to contact and quit trying to strike everyone out... well, they're probably wrong.

Fascinating. More conventional wisdom defenestrated.

Interesting. It looks like the high K pitcher will be roughly about a half pitch/ per inning higer. Over a 200 inning season that is only 100 pitches or roughly one start.

Over a 15 year career thats only 15 extra starts or a half a season. I know this is rough math, but magnifying out these statistics that James provides really does show the fallacy in that theory.

Another little gem from Bill James online. You know how it's taken as gospel that pitchers need to pitch to contact to lower their pitch counts, and therefore go more innings, right? And how young fireballers will try to strike out fewer batters as they mature to save their arms, right?

Well, it's wrong. High strikeout pitchers, on average, throw only a fraction of a pitch per inning more than low-K pitchers:

High K pitchers tend to be more effective than low-K pitchers, so they retire batters more frequently and they get out of innings just as quickly. Also, walk rate is nearly unconnected to strikeout rate in modern baseball. This was probably a bit different in the past, but today there is almost no correlation between K rate and walk rate.

So, next time you hear somebody saying that the only way Daniel Cabrera will become an innings-eater and a real pitcher is to learn to pitch to contact and quit trying to strike everyone out... well, they're probably wrong.

Very interesting.

I can't help but wonder about the methodology. It's not clear to me whether high-K guys are better vs. whether better guys just get more K's.

This is not just a semantic difference. If you look at the IP, there's a huge diff. I think it's highly iffy to lump #1-5 starters together. I'm not saying the guy's wrong, but I wish he'd have used a basis for comparing guys other than "are they in somebody's rotation". I wonder what it would show if he used some reasonable quality-of-performance criteria to compare guys. I fully agree that this is interesting, and I have often wondered about what the data would show. However, I think we can agree that it's rather iffy to say that all SP's are comparable.

What's "striking" (heh heh) to me is how little diff there is between pitches per inning. I think it's safe to say that a group of pitchers who throw nearly 20% more innings than other guys are just flat-out better pitchers. It doesn't surprise me that their K's are higher. I also bet their batters-faced-per-inning is lower too. If you look at these numbers, there are big differences everywhere... except in pitches-per-inning. If the good P's are facing less batters but have the same pitches-per-inning, then their pitches per batter would be higher. Where's the column for pitches-per-batter?

I wonder what it would show if he compared K's and pitches-per-inning among SP's of comparable quality. Maybe it would show the same thing, I don't know, but my hunch is that it would not. I think what he's really comparing here is good-P's vs. bad-P's (as reflected in IP).

ps: Personally, I never thought that DCab should try to hit more bats. I thought everybody wanted him to throw fewer BB's, not fewer K's ;-)

Very interesting.

I can't help but wonder about the methodology. It's not clear to me whether high-K guys are better vs. whether better guys just get more K's.

I would have figured the latter, and I don't see how James took that into account.

Thanks for posting this. I'm still letting it sink in.

First thing that came to mind was (then) Mets pitching coach Mel Stottlemyre vowing to prolong the career of Dwight Gooden following his awe inspiring first two seasons (combined 41-13, 2.00 ERA, 544 K's in 494 2/3 IP) by teaching him how to get outs with one pitch instead of three (i.e. fewer strikeouts).

Well, Doc had a litany of issues that we learned about later, but it was all downhill from there.

Very interesting.

I can't help but wonder about the methodology. It's not clear to me whether high-K guys are better vs. whether better guys just get more K's.

Good point. (Mostly because it's one that I made a while ago in a similar thread in the O's section. ).

I think you need to look at performance in a far more isolated fashion than James did.

I didn't give it a good enough look when I first espied it. And for that I'm ashamed. Nice pick-up on the methodology.

Where's the column for pitches-per-batter?

Sorry about the ugliness of the chart.

At this point I added to the data the numbers of pitches thrown by each pitcher. The pitchers throwing the most pitchers per batter and the fewest pitchers per batter, among the 360 pitchers in the study, were:

            1.  Al Leiter, 2004                   4.33                 1.  Greg Maddux, 2007            3.26           2.  Al Leiter, 2005                    4.25                 2.  Greg Maddux, 2002            3.27               3.  Chris Young, 2006              4.12                 3.  Greg Maddux, 2005            3.31           4.  Chris Young, 2007              4.09                 4.  Carlos Silva, 2006               3.32           5.  Geremi Gonzalez, 2003       4.08                 5.  Carlos Silva, 2004               3.33           6.  Scott Kazmir, 2007 4.07                             6.  Josh Towers, 2005              3.37           7.  Gil Meche, 2006                 4.07                 7.  Jason Johnson, 2005           3.38           8.  Daniel Cabrera, 2006          4.05                 8.  David Wells, 2003               3.38           9.  Gil Meche, 2005                 4.04                 9.  Roy Halladay, 2003 3.39           10.  Matt Cain, 2006                4.04                 10.  Chien-Ming Wang, 2006 3.39

Thus, we realize that throwing a large number of pitches (or a small number of pitches) per batter is a very pitcher-specific trait. Gil Meche has quite ordinary strikeout and walk numbers, nothing remarkable about them. He just uses a ton of pitches to do it. There are other pitchers who have strikeout and walk numbers much higher than Leiter and Chris Young; they just do it with fewer pitches.

I wonder what it would show if he compared K's and pitches-per-inning among SP's of comparable quality. Maybe it would show the same thing, I don't know, but my hunch is that it would not. I think what he's really comparing here is good-P's vs. bad-P's (as reflected in IP).

It appears that there is little correlation between number of pitches thrown per batter and quality of pitcher.

The high strikeout pitchers are better pitchers, in general. I ran some numbers in another thread a week or so ago that showed a decent correlation between number of strikeouts and ERA or FIP ERA. What's even more important in K:BB ratio. That correlated even more strongly to ERA/FIP. It's the strikeouts and the relationship between strikeouts and control that drive how good a pitcher is, to a very large degree.

Another way of looking at this is this chart from the article (last one I'll post, since this is supposed to be little quotes from a subscriber-only piece):

Strikeout pitchers, on average, throw no more pitches per inning than do non-strikeout pitchers, at least within this data. The average number of pitches per inning for the pitchers in these nine groups was:

High Strikeout High Walk 16.72

High Strikeout Medium Walk 16.06

High Strikeout Low Walk 15.16

Medium Strikeout High Walk 16.88

Medium Strikeout Medium Walk 16.03

Medium Strikeout Low Walk 14.97

Low Strikeout High Walk 16.55

Low Strikeout Medium Walk 15.96

Low Strikeout Low Walk 15.27

Pitches per inning vary substantially with walks, but vary with strikeouts only to a small degree. The data above can be re-stated as follows:

High Strikeout Total 15.98

Medium Strikeout Total 15.96

Low Strikeout Total 15.93

High Walk Total 16.72

Medium Walk Total 16.01

Low Walk Total 15.13

Walks drive pitch counts much more than Ks.

Sorry about the ugliness of the chart.

Using the code tags can fix it, but it's a royal pain to use it because you still gotta fiddle with it to make everything line up right.

It appears that there is little correlation between number of pitches thrown per batter and quality of pitcher.

Right. You can throw a lot of strikes... or a lot of balls.

Meanwhile, it only takes 1 pitch to throw an easy-out GB... or a dinger.

So, what do we really wanna know? I'm not exactly sure. I'd think batters-faced-per-inning would be one thing that matters.

I think my wish-list of findings would include pitches-per-retired-batter and pitches-per-batter-who-got-on-base. Which might tell us nothing, or it might tell us something. Sometimes, you just gotta go fishing around some. I imagine that strike/ball ratio (or whatever the right name for that is) separates good guys from mediocre guys. How strongly does that correlate with K's?