Adventures In Heat Maps

The heat maps on our site are pretty amazing. Maybe you noticed that I’ve been having a lot of fun with them, as I wrote a piece on Derek Holland (high fastballs) and Zack Greinke (missing high curves) on the same day last week. Enthusiasm!

They are a tempting thing, these heat maps. Let’s just take a little trip together down Heat Map Lane, why don’t we? I’ve got my Mr. Rogers sweater on, and you’re stuck inside anyway. While Steve Slowinski’s library introduction to the tool is great, it might be worth visually exploring some of the things we can and can’t learn from heat maps.

One thing that becomes clear, quickly, is that it’s hard to say much about control, in a general way, from a heat map. Last year, Jonathan Sanchez had a worse walk rate than any other qualified starter in baseball. Roy Halladay owned the second-best walk rate in baseball. They are generally thought to be on opposite sides of the control spectrum. Looking at the fastball heat maps for Halladay and Sanchez (versus righties), you might not know that was the case. They both have plenty of high-volume clusters (yellow in the heat map) that are outside the strike zone.

What you can tell from this juxtaposition is just as interesting as what you can’t tell, however. For one, it looks like Sanchez likes to aim his fastball outside and high while Halladay prefers to bust the batters in on the hands with his fastball. This is probably the best way to use these heat maps as currently constructed. Because we don’t have the splits by count yet, we can’t tell if that ball that’s three inches outside the zone was thrown in an 0-2 count or in a 3-1 count, which makes a huge difference in intent, and therefore in the pitchers’ ability to command the ball.

Also, you’ll notice that I used 2008 fastballs for the comparison. If you scan up on Halladay’s fastball heatmap page, you might think that he stopped throwing so many fastballs in 2010, and added the mythical cutter. Instead, it’s just a change in classification by the MLB pitch f/x data, and a common wrinkle in the space-time continuum that you’ll notice across many pitch f/x systems. As we perfect our ‘buckets’ and classification systems, we’ll see this disappear, but in the meantime you’ll want to check with the Google to see if a pitching coach or pitcher actually discussed adding or subtracting a pitch.

Let’s head on over to Cliff Lee to see if we can find a change in approach. It seems that we can look at locations to see if a pitcher adjusted and started to pitch differently. Take a look at Lee’s fastballs from 2008 (left) to 2009 (right) and it looks like he’s generally gone from pounding the zone low and outside to bringing them in high and tight:

Cue the whammy sound. What actually happened, upon further inspection, is that MLB pitch f/x took the two-seamers out of his fastball classification. Take a look at Lee’s two-seamer heat maps, and you’ll see that he uses it more on the outer half of the plate for right-handers. So that’s where the missing low-and-outside fastballs went. We must be ever-vigilant.