Baseball Musings
Baseball Musings
February 20, 2006
Strikeouts per 100 Pitches

Rich Lederer suggests that strikeouts per 100 pitches is the best way to judge strikeout pitchers (emphasis in original):

Now, just as K/BF is a better gauge than K/9, strikeouts per total pitches is even better yet. In fact, it is the best one of 'em all. Yes, strikeouts divided by total pitches is the single greatest Defense Independent Pitching Stat out there. It measures dominance and efficiency.

Just as striking out the side in order is preferred over getting all three outs via the K regardless of the number of batters faced, a pitcher who strikes out hitters on three pitches is more effective than those who take five or six to get the job done. By definition, he is missing bats a higher percentage of the time and is also more likely to pitch deeper into games and record a greater number of outs than his counterparts.

I don't agree with this. Striking out a batter is a process. Setting up a batter for a strikeout may easily take five or six pitches. I've seen strikeouts occur because a pitcher got ahead 0-2, then worked the player outside and inside for balls, then threw a pitch that looked like one of the former, but was in fact different.

If you want a pitcher to be efficient, you don't want him striking out a lot of batters. You want a lot of first pitches put in play weakly. But apart from Greg Maddux, there aren't too many hurlers who can do that. Maybe teams should work more on getting their strikeout pitchers up to 120 pitches a game.

Update: From the comments, I believe people are misinterpreting what I wrote in the last paragraph. I'm not advocating a staff of "contact" pitchers over a staff of strikeout pitchers. I'm pointing out that really efficient pitchers get the ball in play early in the count. From 1990 to 1999, Greg Maddux pitched 166 2/3 innings more than any other pitcher in the majors, yet he ranked about 9th or 10th in pitches thrown. Maddux struck out a decent amount of batters, but didn't walk anyone. He also had a lot of balls in play early in the count. With all that, he led the majors in ERA by a good margin (1000 innings pitched).

Now, it could very well be that Maddux's strikeouts per 100 pitches that decade is very good. I don't have access to pitch data for that decade anymore. My point is that there's a tradeoff between efficiency and strikeouts. That if you want a pitcher who is both very efficient and very good, you want Greg Maddux. And a big reason Maddux was both, is that he got batters to put bad pitches into play.

Maddux, however, is an incredible outlier. I'd rather have a pitcher who strikes out a lot of batters. If he doesn't walk too many he'll be efficient enough. So again, while this is a fine way of looking at things, I don't believe it's really telling us something we don't already know.


Posted by David Pinto at 08:07 PM | Pitchers | TrackBack (0)
Comments

sure, contact pitchers can be more efficient than strikeout pitchers, but assuming that in general we prefer k pitchers, and all else being equal, wouldn't we prefer the efficient k pitcher over the inefficient?

Posted by: amos at February 20, 2006 09:31 PM

I'm really not believing what I'm reading here. I've seen teams not have any batters walk and have their fielders make more errors than the opponent and still win the game -- so should we throw out the past 20 years of progress we've made in analyzing this stuff?

Posted by: Tom at February 20, 2006 09:40 PM

I disagree David, on a few fronts. Maddux is acually one of the case examples of inconsistency in batting average on balls in play... which tells me that he's not too special at inducing weak balls in play - if he was, I'd think he'd consistently be a league leader in lowest BABIP. (although I guess velocity data on balls hit in play is available, so I could be wrong... maybe he gives up weaker balls than his peers, but his fielders just stink at turning them into outs).

The thing about a ball in play is that it always has a chance to go for a hit (or an error, or any other non-out). And, of course, a strikeout is an automatic out, every time.

Going back to the idea that an out is the most precious commodity that a lineup has (which seems to be a popular thing to say these days), I'd think a walk would be terrible, since it has the lowest chance of becoming an out (0%), followed by a line drive, then a fly ball, then a grounder, and finally a strikeout (100%). So that's why a K/BB ratio is a good stat - it's got the best outcome on top, and the worst outcome on the bottom, so the higher the better.

Maybe this guy is onto something about pitch count... In dealing with good pitchers, you certainly want them (all else being equal) to be as efficient as possible... that way they can go deeper in games and be "good" for as long as possible. So while I understand how the K/pitch number might be a little misleading due to pitchers setting up batters for the process of a strikeout... it definitely makes sense that, given two pitchers with equal or similar K/BB ratios, I definitely want the guy who takes the fewest pitches to do it.

Posted by: Mike at February 20, 2006 09:44 PM

I don't think people should confuse what is being measured. Strikeouts per 100 pitches is fine if you want to measure the most effective strikeout pitchers. That's it.
The best strikeout pitchers are not always the best pitchers, and they are certainly not necessarily the biggest winners. So strikeouts per 100 pitches shouldn't be considered as a way to measure the best pitchers overall.


Posted by: George S at February 20, 2006 11:16 PM

As for a strikeout being an automatic out; I'd ask Josh Paul and Kevlim Escobar of LAA about that. :P

Posted by: bryan at February 21, 2006 01:16 AM

To read a bit too closely into what Rich Lederer wrote, he calls his stat a measure of "dominance and efficiency."
Those are certainly excellent indicators of what kind of pitcher you have out there. But a pitcher can be dominant in a performance and still lose -- Nolan Ryan being the ultimate example (although he lacked in the efficiency department).
One might argue that strikeouts per 100 pitches is a most realistic barometer of what you're going to get out of a pitcher each start, though.

But David still has a point in noting the blind spot here -- at the major league level, you're going to get most of your strikeouts through getting ahead in the count and then setting up the batter. Three-pitch strikeouts also mean you're giving the batter hittable pitches 100% of the time, and that'll come back to bite you more often than not.

Posted by: James d. at February 21, 2006 02:01 AM

"So strikeouts per 100 pitches shouldn't be considered as a way to measure the best pitchers overall."

Johan Santana, Jake Peavy, Pedro Martinez, Mark Prior, Chris Carpenter, Randy Johnson. Hmm...sounds like it's a pretty damn good metric to me.

You can look in the paper to find out the "best strikeout pitchers". It's true, the best strikeout pitchers are not always the best pitchers, but that's not what Lederer's study is showing. He's showing that the best pitchers are the most efficient of these strikeout pitchers.

And for David's comments that "striking out a batter is a process," well, yes it is. But if you knew the end result, wouldn't you rather have a pitcher who strikes out a batter in three pitchers rather than one who does it in six pitches?

I'll let you have a rotation of guys who put "a lot of first pitches put in play weakly" and I'll take the top five from Lederer's list.

Posted by: Joe at February 21, 2006 02:27 AM

I think the one thing we need to see is which of the two correlates better with overall performance to really judge which is the better of the two metrics. Those who are jumping on the bandwagon without asking whether there is the higher correlation to overall performance are in a way acting like those who believe that sabermetrics is a waste, somewhat blindly forging ahead with their point. I do think it's an interesting look, but until I see that it correlates better with being a good pitcher than K/9, I think I'll use both K/9IP and K/100.

Posted by: Adam B. at February 21, 2006 03:16 AM

If you can only use K as the stat, here are the R-squared results for MLB's last year top 40 ERA pitchers across these metrics:

K/9: 0.0563
K/BF: 0.0984
K/100 Pitches: 0.1298

So, in a vacuum, K/100 pitches is the best of the lot.

But once you start adding other dimensions, you improve your correlation significantly. A quick sample:

K-BB/BF: 0.1694

Or how about:

(K-BB/BF) - ((HR/BF)*8): 0.3998

There's a jump for you! In fact, HR/BF correlates at a 0.2225 rate by itself. So apparently, keeping the ball in the park is the best of all...

Posted by: Dave S. at February 21, 2006 10:12 AM

The correlation was to ERA...

Posted by: Dave S. at February 21, 2006 10:17 AM

Dave Pinto,

Maddux had a k/100P of 5.9 for his 9 years with Atlanta. That would have placed him in the top 10 of the league last year. If you had k/100P + HR rate, you'd have the reason Maddux was so good.

K/100p picks up BB rate really well, and therefore, it picks up some of Maddux's greatness at not walking people, even if it doesn't pick all of it up.

Posted by: joelw at February 21, 2006 11:30 AM

Someone wrote earlier that a pitcher with a lot of strikeouts isn't necessarily a good/great pitcher. I would disagree. Strikeouts, if one just looks at the league leaders over the years, seem to be a pretty good indicator of quality. I mean, aside from the occasional Bobby Witt, generally the league leader in Ks is a good pitcher overall. While the "contact" guys might have a good year here and there - like Tewksbury, etc. - but don't maintain dominance for extended periods, generally. Maddux isn't really a "contact" pitcher - in his prime he was racking up 160-190 Ks a season.
Strikeout leaders:
NL-AL
2005 - Peavy-Santana
04-Johnson-Santana
03-Wood-Loaiza
02-Randy-Pedro
01-Randy-Nomo
00-Randy-Pedro
99-Randy-Pedro
98-Schilling-Clemens
97-Schilling-Clemens
96-Smoltz-Clemens

You get the picture - only Loaiza and Nomo could be considered less than elite. Strikeouts are the key to pitcher quality.

Posted by: david at February 21, 2006 09:24 PM
Post a comment









Remember personal info?