Pay More Attention to Swings and Misses

As an analyst, one of the aspects that I am obsessed with is boiling down seemingly complex questions into the simplest possible situation. Take evaluating pitches for example. John Walsh of the Hardball Times has an excellent method for identifying and ranking pitch types but a data request from my friend Jeff Sullivan got me thinking about a simpler model.

What is the best result for a single pitch, from a pitcher’s point of view? Clearly, the pitcher prefers a strike to a ball. An out would be best, but since we’re talking about a single pitch, an out means a ball in play which in turn means the possibility for a whole range of other outcomes.

A principle that I have been harping on for a few years now is the benefit in splitting up pitch results at a level beyond ball, strike and in play. In particular, I tend to categorize pitches as one of the following: ball, intentional ball, called strike, swinging strike, foul or in play. Why do I think it’s important to split things up like that? Because of graphs like these:

That’s strikeout rate on the vertical axis and league normalized swinging strike and called strike rates on the horizontals.

Going back to the question then, would anyone disagree that a swinging strike is the best overall pitch outcome for a pitcher? Not only does it result in the best possible singular outcome (a strike), but it adds a lot more information about the pitcher’s ability to get strikeouts, the best possible outcome of an at bat for the pitcher.

Looking at fastballs only and the percentage of them that a pitcher gets a swing and a miss on, there’s some names in interesting places. It’s probably no surprise that Scott Kazmir and Rich Harden were at or near the league best in 2008, but would you have guessed John Danks and Micah Owings would be right with them?

It’s probably not surprising that Dallas Braden had the lowest rate of fastball swinging strikes amongst starting pitchers last year given his typical 87-88 mph velocity, but would you thought Bobby Jenks and his mid-90s fastball would be near the bottom for relievers at a rate roughly half that of Ramon Ramirez? In fact, Bobby Jenks saw a significant decline in the amount of swings and misses he managed across all pitches in 2008, something to watch out for in 2009.

There’s no end to the amount of interesting (to me) data that can be looked at concerning pitchers.

Print This Post

Matthew Carruth is a software engineer who has been fascinated with baseball statistics since age five. When not dissecting baseball, he is watching hockey or playing soccer.

9 Responses to “Pay More Attention to Swings and Misses”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. Crane says:

    I knew Micah Owings was really cool. I totally knew it.

    Vote -1 Vote +1

  2. Fresh Hops says:


    Out of curiosity I grabbed all pitchers with 15 or more starts last season and took a look at swinging strike rates and looking strike rates in a pretty simple way:
    The average called strike rate was 17.9% with a maximum value of 25.3% and minimum of 14.5% (range of 10.8) and a standard deviation of 1.6.

    The average swinging strike rate was 8.4% with a max of 15.1% and minimum of 3.7% (range 11.4) and standard deviation of 1.9.

    There’s greater dispersion in swinging strike rates. This suggests that it’s an ability a pitcher has more ability to control, which one might speculate is true from the get go: a pitcher can only throw so many strikes in the zone that batters don’t swing at.

    Thanks for the link to the Walsh article. that was nice stuff.

    Vote -1 Vote +1

  3. Jonathan says:

    This was a nice post, but I’m not sure I understand why a strikeout is the best possible outcome for a pitcher. Isn’t this going to depend on context? Here are two examples, one local and one global in character. Local. With one out and a runner on first, the pitcher is pitching to the opposing pitcher (or maybe to the number two hitter or something similar). Surely, it would be better to induce an inning-ending double-play than to strike out the batter. Global. The pitcher is old and/or has lower-than-average endurance. For such a pitcher, it seems the best thing would be to minimize the expected number of pitches thrown to the opposing hitter, say aiming for something less than three. Such an approach might make the difference between pitching five innings and pitching six innings. But doing this means striking out fewer batters. I’m happy to be shown what I’m missing here.

    Vote -1 Vote +1

    • Matt H. says:

      Because balls in play bring in things the pitcher can’t control like defense.

      Vote -1 Vote +1

    • Fresh Hops says:

      Efficiency studies on pitchers have shown that the most efficient ones are the ones with high strike out and low walk rates. (Efficiency is just pitches per out.) It takes more pitches to strike a batter out than to put a ball in play; however, about 1/3 of balls in play are hits, so that means the pitcher has to face another batter. Equally importantly, if the ball in play is a hit, you have a runner on base with the opportunity to score (or two of them in the GIDP case.) Take a look at pitch to contact guys (Carlos Silva) and compare with some guys that do 7K/9 and 2.5BB/9. The latter are allow fewer runs and don’t throw more pitches per game. So I think the global case is mistaken.

      The local example is important. You’re certainly right that ending the inning with GIDP is better than a strike out. But the question is whether the pitcher has much control over that outcome. The received view is that a pitcher effects Ks and BBs to a high degree, some has some control over whether batted balls are GB or ball in the air, and little control over the rest. If that’s right (and it has some pretty good evidence going for it), there’s probably not much a pitcher can do to induce GIDP over and above his ability to produce a GB. Even the best GB pitchers can’t induce GB on more than about 65% of batted balls, so that suggests a pretty small ability to produce GIDP.

      Vote -1 Vote +1

  4. thumble says:

    John Walsh has 4 PA? (Related Batters info box at the top right)

    Maybe John could give us a breakdown by pitch type and sequence, although the .000 wOBA does not look promising.

    Vote -1 Vote +1

  5. Bryan says:

    Just because called strike percentage has a low correlation with K% doesn’t mean that only swinging strikes matter. All the low correlation shows us is that as a population, pitchers show no trend between the two. Since there are many different types of pitchers we might expect many different types of behaviors. All of those lumped together might drown out any local correlation.

    In particular, called strike % on breaking balls probably correlates to strikeouts. How many times have you seen a batter freeze only to see the curve drop straight through the strike zone? It would really surprise me if good pitchers don’t get more called strikes.

    The outcome of a swinging strike and a called strike are exactly the same all the time. The only reason I could see any difference between the two is if most strikeouts ended with a disproportionately high amount of swinging strikes. It would be interesting if this was the case. That could also explain the very high correlation.

    Vote -1 Vote +1

  6. MattS says:

    This is definitely interesting stuff. I think that these kinds of things are far more useful for looking at rookies or players with very little data available on them, rather than being helpful for players with other data. Swinging strikes, called strikes, all types of things like that are very noisy signals of what we’re actually trying to predict– outcomes. I did a piece recently at StatSpeak, where I discussed including stats like this in regressions to predict things.

    The pitcher one is not actually listed, but it followed the same pattern– when I regressed strikeout rate on the previous year’s strikeout rate and the previous year’s contact% (as listed by fangraphs), the contact rate out significant and predictably so– less contact (and more swings and misses) meant that the pitcher was likely to strikeout more people the following year than the earlier year’s strikeout rate would predict alone. However, a second year of strikeout data left contact% very statistically insignificant. Have a look at the article.

    That’s not to downplay this article– to the contrary, it’s to focus where it’s useful and that is looking at young players. The old ones have enough data that a noisy signal of skill won’t help.

    Vote -1 Vote +1