Labwork: Brian Wilson FB Location vs. LHB

Published by on February 1, 2011 | 9 Responses

First: if you’re a graph nerd like me, and you haven’t been reading The Prince of Slides blog, you need to go check it out now. You’ll notice that it’s been added to the link section on your right under ‘Friends of BCB’. Millsy has been on an absolute tear with his sab-R-metrics series — an introduction/primer to using R with baseball data such as PITCHf/x. I’m an Excel junkie, but I’ve been trying to learn R for the past few months and the sab-R-metrics posts are a godsend. Thanks, Millsy!

Second: heatmaps seem to be all the rage right now in PFX analysis. They’re interesting, and fun to look at. In fact, FanGraphs just announced that they are adding them to their site as a feature (read: Customizable Heatmaps!). However, being someone who likes to control the fine details of what they’re doing, I’m still very interested in generating my own heatmaps.

With the help from the sab-R-metrics series, I’ve been working in R to try and display Brian Wilson’s FB location by batter-handedness. As a sample, here’s a map of Wilson’s FB location versus left-handed betters from 2009 – 2010.

(Graph Note: remember, it’s based only on location and not any sort of run value. The greater the density of that pitch, in that area, the darker the heatmap is.)

It appears (and I assume it’s not that uncommon) that Brian Wilson pumps his heat away from left-handed batters. It seems like a good strategy for a pitcher that can run his fastball into the mid-to-upper 90s. As the saying goes, “Make them beat you the other way”. However, you’ll notice that my plot is slightly off-centered. That’s been my biggest issue with R. I’ve tried plotting without axes with similar results. I’ve also tried setting the axes parameters with ‘xlim’ and ‘ylim’ but the results were unfavorable. I probably don’t have enough data points, but I’d love to see if Wilson changes his approach when pitching in San Francisco. AT&T Park is notoriously tough on LHBs and I wouldn’t be surprised to see Wilson throw his heater down the pipe more at home.

Thoughts for another post, I guess. I’m still working out the problems with R, but if anyone that reads this site is an R user, I would love to hear any suggestions on how to fix my axes issues.

—-

UPDATE:

Not perfect, but probably the best I can do right now.

(Click to enlarge)

Thanks go out to Millsy at The Prince of Slides for his help. I couldn’t have done it without his excellent site.

Over the past two years, with the fastball, Wilson has worked left-handed batters away and pitched more to the heart of the plate against right-handed batters. There may have been some confusion in the comments section on how to read these graphs. The vantage point — like 99% of PFX graphs — is from the catcher’s POV. Just envision that you’re the catcher, sitting behind the plate, looking out towards the pitcher.

Article written by

9 comments
  Livefyre
Newest | Oldest
elliotjorgensen
elliotjorgensen

My favorite spots on the heat map are to the positive side (greater than 1) and up high (around 5). A 98 mph fastball at a batter's chin can really open up the outside corner for the following pitch. And I do agree that Wilson is more likely to pitch inside at home, with death valley, than on the road. Especially when compared to a place like Enron statdium (hah! Enron).

Ben M
Ben M

Okay, this must be a stupid question since nobody else has mentioned it, but doesn't the heat map show that he throws inside to lefties? Or is it taken from the batter's point of view?

Chris Quick
Chris Quick

Sorry, I should be more clear: graph is from the catcher's POV -- ie: sitting behind the plate

Millsy
Millsy

Just tried the white "low density" palette and it works great. You can make your own palette with whatever colors you want. I have a red palette working fine, but going to fiddle with some more colors and I'll post something later today.

Millsy
Millsy

I know I've tried the xlim command. However, it's not necessarily the axes that are a concern (I can set those just fine). The function used to create the plots automatically chooses the range to smooth, and setting axes limits directly unfortunately does not set the range for the smoothing. By simply setting axes, it ends up leaving a bunch of blank area in the background, and the standard "bg" command won't fill it in. One suggestion I do have: just set your palette to have the 'lowest density' areas to be white. That way you won't be able to tell it's a bit wonky. I'll have a post on this soon. Thanks again for the link back.

Chris Quick
Chris Quick

>> One suggestion I do have: just set your palette to have the ‘lowest density’ areas to be white. That way you won’t be able to tell it’s a bit wonky. I’ll have a post on this soon. Thanks again for the link back. Heh. That's an interesting way to work around it. Maybe I'll try that later. Millsy is right about the smoothing function. It chooses the range based on data. You can change the axes, but you just get white area instead.

Chris Quick
Chris Quick

I've got a really 'hacky' idea on how to make it centered. I'll try it out later and see what happens. Thanks for stopping by!

TheFedman2002
TheFedman2002

My buddy uses R frequently, he said the following: "set xlim=c(-3,3) or something like that. There's also some command like par(pin=c(width,height))"

Millsy
Millsy

Thanks for the link back, Chris. The plots look good. Interesting to see Brian Wilson pitch Mariano Rivera-esque-y. If I figure anything out with the smoothing range (so it can be centered) I'll let you know.