Why Swinging Strikes Are So Important to Pitching
After posting my article on Pitch Outcome Game Score and tweeting it out, renowned sabermetrician Tom Tango asked for further proof that swinging strikes are so much more important than other types of strikes. I’ll attempt to provide this by going into more detail on the process to determine pitch run values for poGS, and show in different terms how a swinging strike is so indicative of a successful pitcher.
To determine run values for individual pitch outcomes, I connected every pitch in the Pitchf/x database (downloadable at Baseball Heat Maps) with the result of the plate appearance it took place in. If a five-pitch plate appearance ended in a double, each of the pitches was assigned a double. Then I used the standard set of linear weights described here by Tango to assign each of the pitches a specific run value. A double is worth .76 runs, so each pitch in the plate appearance is assessed a value of .76 runs. This may seem to reward bad pitches that precede good outcomes and unfairly penalize good pitches that precede bad outcomes, but over millions of pitches, the cream should rise to the top. Here is a comprehensive list of the pitch outcome/run value mappings. Some of the lesser-used outcomes may look strange and inadequate, but they are what Pitchf/x gives us.
Run Value Above Average
|In play, run(s)||132511||0.666|
|In play, no out||251832||0.469|
|Hit By Pitch||13197||0.278|
|Ball In Dirt||102945||0.0315|
|Foul (Runner Going)||21936||0.0153|
|In play, out(s)||719413||-0.177|
|Swinging Strike (Blocked)||33491||-0.270|
These pitch outcomes can be categorized into five groups: ball, called strike, foul, swinging strike, ball in play. For the purposes of poGS, I also combined called strikes and fouls, as I found that the difference between them wasn’t large enough to separate in a metric that needed to be easy to calculate.
The run values that I received after combining are:
- Swinging Strike: -0.169
- Called Strike/Foul: -0.040
- Ball: 0.044
- Ball in Play: 0.071
Knowing that a particular pitch is a swinging strike should lower your expectation of the pitch’s plate appearance to -0.169 runs below average. If I instead knew that the pitch was put into play, my expectation of the plate appearance should rise by 0.071 runs. Swinging strikes, by this methodology, are more impactful on plate appearances than called or fouled strikes are.
What Does Swinging Strike Rate Tell Us About a Pitcher?
The other way to analyze the impact of swinging strikes is to look at pitchers’ rates of the different types of strikes and compare them to FIP, which measures overall pitching effectiveness. A strikeout can occur on either a swinging strike or a called strike, so if we’re just interested in the result of a plate appearance or game, the type of strike doesn’t really matter. But swinging strikes may tell us more about a pitcher’s ability to limit runs than called or fouled strikes.
I compiled the swinging, called and fouled strike rates of 40+ inning seasons from 2011 to 2014, as well as the corresponding FIPs. The data counts partial seasons separately (players who change teams mid-season can be counted multiple times), since it came from Baseball Reference, but this shouldn’t affect our findings.
Here is swinging strikes per pitch correlated with FIP. We get a .3171 R^2 value.
Next, here is the same for looking strikes per pitch and FIP. We get a 0.0464 R^2 value.
And finally, fouls per pitch and FIP. A 0.02 R^2 value.
So it certainly appears that swinging strikes tell us a lot about a pitcher’s ability to limit runs, in terms of FIP. Fouled and called strikes really don’t, as their coefficients of determination with FIP are nearly zero. A pitcher can’t consistently depend on fouled and called strikes carrying his production. These correlations seem to conform with what I found in my poGS research. A swinging strike tells us good stuff about a pitcher that other strikes do not.