About the blog:
Clown Hypothesis is a place for empirically-based sports commentary, research, and analysis, mostly about baseball and hockey. (The name is inspired by Bryce Harper’s memorable response to a dumb question.)
About the author:
Frank Firke is a lifelong Chicago sports fan and math nerd. He works as an analyst for an economic consulting firm in New York City. You can find his tweets @ClownHypothesis.
That's a Clown Hypothesis, Bro! 
I stumbled upon your blog while searching for an answer to an (it turns out) obscure question. I found your article on pitcher’s diversity in # of pitches and whether it correlated with the number of pitches thrown total.
The article was excellent, but wasn’t exactly what I was looking to find. My question is whether, as a batter, you can gain information about an upcoming pitch when you see the pitcher shaking off the catcher 1, 2, 3 times etc.
For example, in situations where a pitcher shakes off the catcher 2 times, is the pitch that will be coming more likely to be a fastball? I could see it possibly being pitcher dependent (in which case the data samples would likely be too small except for multi-years 200+ inning guys) , but then I could also see it being a general thing–or maybe the distribution would be identical to pitches where no shake-offs occur.
I know pitch selection data is kept, but are # of Shake-offs and what types of pitches were shook off kept? I guess if someone is on second, there’s no way of knowing which pitches were shook off though–but still the pitch that was ultimately thrown would be known .
Is this type of data kept? What’s your gut reaction? In my anecdotal confirmation-biased experience as a batter, I feel like if a pitcher is shaking off pitches, a challenge-the-batter-even-though-it-might-not-be-prudent fastball of whatever the pitcher’s specialty is often on it’s way.
But then, confirmation bias is a bitch and I am very open to the likelyhood I’m dead wrong about this. Any feedback would be appreciated! -Paul
Hi Paul,
Glad you liked the piece. As far as shakeoffs go, I suspect that’s going to be (near-)impossible to assess quantitatively. I don’t know of any place that tracks that, and it seems like the sort of thing that would be very difficult to track—you can’t use the broadcast because it cuts away too frequently, and I’m struggling to think of a better tactic than parking an intern with a notebook/tablet in the scout seats (or getting a camera that doesn’t cutaway and using that).
My intuition (take this with a big grain of salt) is that a shake-off is only going to be informative in the context of the pitcher, the batter, and how the pitcher’s feeling that day, and so it would be hard to take an aggregate effect and apply it to an individual situation, even probabilistically speaking. (This is all assuming the behavior has any predictive behavior, which wouldn’t surprise me too much but certainly isn’t guaranteed.) Basically, I imagine that disagreeing with the sign means something very different when it’s Chris Sale than it does when it’s Hector Noesi, and it means something different when pitching to Mike Trout compared to having Jose Molina in the box.
I suspect that you’re probably right about it being pitcher specific–I forgot to take into account that my own experience batting is not at a professional level (or even a high minor-league level) and that the battery has scouting reports on each individual batter. This is qualitatively different than my experience at the high school and collegiate level, where scouting reports are done on the pitchers but not so much the other way around.
Nevertheless as the catchers often getting signs from the dugout or at least following a determined plan, I still have a feeling that shake-offs might well be indicative of the pitcher being less prudent than the catcher/dugout, leading not necessarily to him tipping his next pitch–but possibly skewing his normal behavior in one direction.
Coming back to reality though.. the one thing that you bring up that I think is probably a huge strike against my idea is the fact that this data is not tracked. There are many people far smarter than I working on getting ANY possible edge, so if there was something substantive here it seems more likely that this data would somehow be tracked.
Anyhow, thank you so much for your feedback. I’ve had this idea bouncing around my head for a while but none of my friends or former teammates are interested in applying statistics to baseball. It doesn’t surprise me that even today there’s still certain major league head offices who pay only lip service to sabremetrics–oh well their loss.
Thank you again for your input, -Paul