Are Swinging Strikes Better Than Called Strikes?
Everything you know to be true in your heart but hasn’t been proven by stats is worth hanging on to, even if just a little bit, and privately. The stats may catch up some day. This isn’t to say that all conventional wisdom is correct. This is to say that all “statistically-proven” wisdom is not always going to continue to be true.
Take swinging strikes, called strikes, and Vance Worley.
Vance Worley blew up in 2011. He struck out more batters on a rate basis than he ever had in the minor leagues. He did it with one of the worst swinging strike rates among starters that year. He did it with called strikes — he was fifth among starters with at least 2000 pitches that year. He did it with style, as you can see thanks to Zoo With Roy:
As 2012 approached, I was tasked with figuring out his fantasy value for the upcoming season. I had a personal preference for the swinging strike. To me, there’s no cleaner statistical happening in baseball — that the batter swung and missed is irrefutable. And the swinging strike as a moment is both triumphant and despondent, all in at once. It renders a one-nothing August game watchable. It’s beautiful.
But there were stats, and research, that outweighed my love for the swinging strike. Matt Swartz had found that the year-to-year correlation for swinging strikes (.77) was a little better than called strikes (.59), but he also found that swinging strikes didn’t improve our projections for strikeout rate much.
So I shrugged. I said Worley wasn’t going to be as good, probably, because his minor league stats weren’t great, and his swinging strikes weren’t either. But I did mention that called strikes were decently correlated year-to-year, and he was returning to his same team. All of this in a blog post for the now-defunct Getting Blanked blog that The Score has removed from the internets.
Whatever, the point was that context was huge. Because Worley’s followup wasn’t great, but he still managed to be close to league average despite a terrible swinging strike rate. Once again, he did well in called strikes — 13th this time among pitchers with 2000 pitches. I was right and I was wrong.
Worley threw to Carlos Ruiz those two years. According to StatCorner, Ruiz was a below scratch framer in 2011 (-0.7 calls per game) and above in 2012 (+0.4 calls per game), but never a league leader on either side. Worley also threw in the National League, to certain umpires, in certain parks. All of these things matter to called strikes.
When he moved to Minnesota, it all fell apart. The bad swinging strike rate got worse — he got fewer swinging strikes than anyone in baseball that year — but also the called strikes disappeared. He was 223rd among pitchers with at least 40 innings that year, with a 31.6% called strike rate, down from 36.5% the year before. He fell from atop the leaderboards to below league average (32.9%) that year.
He was throwing mostly to Joe Mauer then, and Mauer was a scratch framer almost literally that year (-0.01 calls per game). So he could’ve been called an upgrade, at least from 2012’s version of Carlos Ruiz. But that’s not the point, the point is that the entire context needed to be taken into consideration, and Worley’s context had changed radically.
Well, two guys have been doing some work on the framing side that has focused mostly on this context. Harry Pavlidis and Jonathan Judge recently debuted CSAA, or Called Strikes Above Average, and it’s great. By taking into consideration park, league, umpire, pitcher, and catcher context, they hope to really hone in on the framing skill of the catcher. It’s a real advancement of the conversation.
Nestled within this work is a CSAA for pitchers, naturally. Pitchers must have some skill at getting called strikes, and since they did all the work quantifying the context, it was easy for them to run the year to year spearman correlations (r) for pitchers’ CSAA:
.@harrypav @enosarris
all pitchers (1+BF): .38
typical bulllpen (170+BF): .46
qualified starter (660+BF): .50
— Jonathan Judge (@bachlaw) March 27, 2015
Now we see that much of the year-to-year strength of called strikes in pitchers is due to the catcher and park influence. If you hone in on just how well a pitcher can get called strikes year to year, it’s not a very strong skill. It’s just a little bit stickier, year to year, than Home Runs Allowed, making it a modestly sticky skill in the pantheon of pitcher skills.
There’s still work to be done. As Judge points out, they haven’t specifically looked at swinging strikes with the context accounted for. But he agreed that there was less likelihood that context contributed as much to a swinging strike.
And really, that’s the whole point. There’s always more work to be done, when it comes to figuring out mechanics, best practices, or statistical truths. Nobody has it all figured out.
With a phone full of pictures of pitchers' fingers, strange beers, and his two toddler sons, Eno Sarris can be found at the ballpark or a brewery most days. Read him here, writing about the A's or Giants at The Athletic, or about beer at October. Follow him on Twitter @enosarris if you can handle the sandwiches and inanity.
An article on this topic was done a week and a half ago:
http://www.breakingblue.ca/2015/03/22/why-swinging-strikes-are-so-important-to-pitching/
Note that a subsequent discussion on Tango’s forum pretty much debunked the linear weights value difference between called and swinging strikes – there might be a tiny difference, but not a huge one. The regression and other bits probably still hold, though.
http://tangotiger.com/index.php/site/comments/are-swinging-strikes-more-costly-than-called-strikes#comments
I’m not really writing about the value of each in the game, unless that’s a reference to the piece I haven’t read yet.
Right, I was trying to say that the Breaking Blue article isn’t necessarily relevant to this one, because it is factually flawed.
Yeah, it looks like I jumped the gun in posting that link. The headlines are similar, but the content is very different. Apologies.
Wait, isn’t CSAA controlling for pitch location? Meaning, what pitchers can get called strikes that they shouldn’t be getting. But if some guys just throw it in the zone more often, they’re going to get called strikes more often, and that isn’t going to show up in CSAA, because that’s not what that statistic is meant to be doing. Taking that out and saying “look, called strikes aren’t that much of a repeatable skill” seems like a bit of a misinterpretation.
I’ll have to defer to the guys that did the research on this question, but they were aware of how I was going to use it.
CSAA does control for pitch location. So pitchers with above-average CSAA are getting more strikes than other pitchers get for throwing in the same locations.
To see that a pitcher’s CSAA is just a bit stickier from year-to-year than BABIP is not really correct based on the link you provided. For the 11 years of 2002-2012, BABIP has a correlation of 0.235 with a 140 innings pitched minimum. CSAA has a correlation of 0.50 with a 660 batters faced minimum. That’s a pretty meaningful difference in my opinion.
I missed the 140 IP thing, my bad. Updated.
Is Judge’s tweet showing R or R-squared?
r
Shouldn’t that be 2012? (Great stuff otherwise!)
sigh
Heres just a thought…if a catcher is overall an average pitch framer, that doesn’t necessarily mean he is an average pitch framer for every pitcher he catches, does it (even accounting for noise)?
If not, then Worley throwing to an overall average pitch framing catcher may not necessarily mean that the catcher was an average pitch framer for Worley specifically.
Likewise, Worley switching from an overall average pitch framing catcher to a different overall average pitch framing could mean that Catcher A was above average specifically for Worley and Catcher B was below average specifically for Worley, resulting in a substantial drop in called strikes.
Just posing the question.
As the impact of pitch framing becomes more an more understood, is there a “matching effect” context that needs to be taken into consideration?
The main skill you are trying to find in called strikes year to year is “How often does a pitcher put the ball on the edge, or just off the edge of the strike zone”. CSAA only looks at the rate in which these pitches are called strikes, not the frequency in which they are thrown and not swung at.
This would be like comparing whiff % of two pitchers, but only on split fingers with location X, and break Y, while ignoring the frequency with which they can pull off that pitch. You might find that a AAA reliever has a similar whiff% as Tanaka on his nastiest splitters, but ignore the fact that he hangs a much larger percentage of his attempted splits.
This article without the “Worley Bird”: A
This article with the “Worley Bird”: A+++++++++
I wonder if all of this research will eventually lead to a version of WAR that values pitchers on a per-pitch basis rather than a per-PA one. It’s always struck me as a bit off that pitchers only get credit for the run value of strikes when they happen to sequence three of them together in the same plate appearance.