Anderson Used to Shift His Position on the Mound
We here at Fangraphs are big Brett Anderson fans: Dave Cameron has a huge crush on him; I fawned over his crazy slider; Marc Hulet saw him as the AL ROY favorite last August and then was devastated when he missed out on him in his fantasy auction; and, RJ Anderson and David Golebiewski see him as one of the game’s top starters. So when a commenter to my post last Thursday on Sheets shifting his position on the mound noted that Brett Anderson used to do this too, I jumped at the opportunity to check it out.
In this image I plot the histogram of his horizontal release points versus RHBs (solid) and LHBs (dash). In April and May he had two clear release points: one to RHBs and another shifted to the extreme arm-side of the mound to LHBs. But starting in June and continuing for the rest of the season — and in his starts so far this season — he uses a single location on the mound, the one he had previously used just to RHBs. On the right side of the graph it shows the number of Ks and BBs he had that game. (There is no pitchf/x data for his April 28th start.)
We all know that correlation does not equal causation — so we cannot say that Anderson started doing better because he stopped using two different positions on the mound — but it is interesting that at the same time he dropped the second position on the mound he also started pitching much better (FIP of 5.88 in April and May versus a FIP of 3.04 from June on).
31
Maybe you meant to say correlation does not equal causation?
Yeah, thanks.
Dave,
I believe his velocity spiked right around the time too. I do not think those two are related, however I think his success can probably be traced back to both events.
Mike,
Yeah his velocity did spike around then. I think that probably played a bigger role than this shifting release point stuff, but I think most people are aware of the velocity stuff and wanted to show the release point.
Nice article Dave, glad I brought it up. It’s nice to see that data with the K’s and BB’s next to it.
So it would seem the moving around on the rubber is a philosophy of the A’s? I would move around for my 2 seam and cut fastballs depending on what handed batter I was facing and what I wanted that particular pitch to do (I pitched to both sides of the plate with both pitches though I generally worked inside to both types of hitters).
Forgot to make my point, did he (and Sheets) only move for lefty/righty or were certain pitch types involved?
No, I am pretty sure both of them did it will all their pitches.
I don’t think so, because Sheets was doing it with the Brewers too, long before he got to Oakland.
Is there any correlation with the data you’ve shown and a pitcher’s likelihood in sustaining a throwing arm injury? Sounds a bit far-fetched (and perhaps a tad irrelevant) but the main reason why avoided Anderson in all drafts were injury concerns relating to his innings jump and pitch selection.
2007 – 120 innings
2008 – 105 innings + whatever he did for team america
2009 – 175 innings
Doesn’t seem too bad to me. They took it easy on him as much as they could last year. He only pitched a few innings in the start after his shutout in Fenway, and they went to a six man rotation for the last month.
…and now he’s got elbow problems. Great.
Sweet picture. Honestly. I have no idea how you even created it.
agreed. very nice visual. I almost don’t even have to read the article.
Thanks guys, the graph is inspired by the Tufte-promoted idea of the small multiple.
Fantastic. Idea: maybe look into how much individual pitchers shift on the mound depending on handedness, and how it affects their platoon splits. I’ll take it if you don’t.
such a nice graphic!
I love all your PITCHf/x work and I’m happy to see that you keep inventing new ways to use the data. We’re only scratching the surface of what can be done, but we’re digging at a steady pace thanks to people like you. Kudos.
Kyle,
Thanks I appreciate the comment.
As Randall Munroe of xkcd puts it, “Correlation doesn’t imply causation, but it does waggle its eyebrows suggestively and gesture furtively while mouthing ‘look over there.’” (http://xkcd.com/552/)
He seems not to have changed vs RH’s. Any before/after data showing performance changes vs LH’s?
That is like the most incredible graph I have ever season.
The type of thing I’d like to see become standard for stuff like this is to attach a p-value (and/or confidence interval for the difference between before and after switch). In this case, it’s a pretty simple two-proportion z-test to do for either k% or bb%. Obviously, correlation doesn’t equal causation, but that doesn’t mean we shouldn’t try to quantify the correlation and its magnitude. You could also put his velocity by the chart for people to eyeball the respective correlations.
Awesome article overall, though – this is the type of stuff I really like to see, analysis of in-game decision-making on the player level, with potentially relevant practical implications. I think this is the frontier we’ll start pushing with pitching analysis, since there’s a lot more optimization possible there given that the pitcher is unique on the field in the impact of his decision-making and the control he has over it.
Stuff like this keeps me coming back to this site. Love the graph.
My favorite thing about this article is that it was inspired by a writer reading an interesting comment. And while fangraphs was already better than espn and the like, that makes it so so much better.
Thanks again guys for being pretty fantastic.