Thursday, April 7, 2011

How To Read Pitch F/X Data

NOTE: This is not intended to be an authoritative manual on Pitch F/X data or data manipulation. I'm unqualified to execute such instruction. Instead, this article aims to make Pitch F/X final products (charts and plots and such) accessible and digestible to average fans.

What is Pitch F/X?
The history of Pitch F/X -- sometimes called Pitchf/x or Pitch f/x -- is actually quite interesting, a history involving glowing hockey pucks and yellow first-down lines. I'm not sure where one could find the best-written history on the matter, but I this Bloomberg article has a dandy cliff-notes version.

Pitch F/X is basically three dimensional data acquired from special cameras set up in each major league stadium (and the minor league/spring training stadium in Surprise, Arizona). The cameras are able to capture (1) the exact locations where a ball crosses the plate, (2) the movement or spin of the pitch, (2) the velocity of the pitch, and (3) the pitcher's release point (the exact angle wherein they let go of the ball).

In other words, Pitch F/X data is a nearly all-in-one scouting report for pitchers, as well as a highly valuable tool for how pitchers are attacking certain hitters. It was originally created (I believe) to help evaluate umpires, and then quickly found use in television broadcasts (where it is now widely used).

Where Do I Find This Data?
Well, the beauty of Pitch F/X data is that it's everywhere. Sportsvision, the company in charge of all this insanity, release the data for free to the public. This has created a myriad of Pitch F/X depositories, each offering some different strengths and weaknesses. Here are three I tend to use a lot:

FanGraphs: The site FanGraphs (for which I now write) started offering Pitch F/X on each player's page. I don't believe you can find data on umpires or across the league or anything, but you can easily create pretty heat maps and charts.

Joe Lefkowitz's Pitch F/X Tool: Not nearly as pretty as FanGraphs, but as complete and thorough as most anyone could want. Also, you can download the raw data and make it pretty yourself (if you're talented enough).

Brook's Baseball: Brook's has a different search method than the other two and can be very useful for examining a single game or at bat.

How to Understand It
Well, each of these three sites offers something very different; also, I imagine there's a plethora of other sites I haven't encountered (or forget at the moment). Nonetheless, with Pitch F/X analyses, we usually see something like this:


This chart shows everything the catcher sees. Typically Pitch F/X charts are from the catcher's perspective, not the pitcher's -- but not always. The box in the middle is a normalized strike zone; the strike zone needs to reflect the differences in players' heights.

However, this is not really empirically digestible or even really pretty, but what it is is Brook's Baseball's full report on Carlos Zambrano's 2008 no-hitter. A chart like this might be used by a rookie sabermagician, but a vet would know there's too much info to digest.

If we head to Fangraphs, we can grab a chart like this:


It's far more pretty, but it's also telling us less and telling us something different. The previous chart just plotted every pitch Zambrano threw with respect to the strike zone (and also the number pitch in the at-bat). This one just shows the movement on each pitch.

The FanGraph's chart doesn't show us any strike zone. Rather, it show how the pitches' trajectories -- how they moved once they were in-air.

"So fastballs move up and to the left?"

No, not quite. It's more of a calibration issue than anything. The fastballs are nearly straight, considering Zambrano's arm slot.

Once we recognize the FA pitches (fastballs) are more-or-less straight, we can then interpret the movement of the other pitches. The above chart doesn't show speed, but we can see the Gameday Algorithm (which classifies pitches) seemed to be mistaking Carlos Zambrano's sinking fastball (or sinker) with his regular, four-seem fastball (FA).

Notice how there's a cluster of fastballs nearly touching the changeups. These fastballs moved vertically lower than the majority of the other FAs, and further left than the changes -- i.e. the moved just like a typical sinker.

If we want to be sure these were sinkers, we can always refer to a chart that combines movement and velocity. Because the sinkers had different vertical and horizontal movement, we can use either dimension to differentiate them:


We can see here how the cluster of fastballs moving to the left were indeed far faster than the changeups -- which were similar in movement. Also, we can now notice a few mis-classified sliders (the two red dots lost amidst the orange changeups).

So, now that we've identified some of the errors in the Gameday Algorithm (errors that have probably been fixed by now), we can delve even deeper into Zambrano's amazing game. We can begin to visualize exactly what tormented those Astro's hitters that day, how exactly he was able to keep hitters off-balance.

And basically, it comes down to this: Carlos was throwing four pitches (his four-seem fastball, sinking fastball, slider, and change) all around the zone (as we can see from the Brook's Baseball chart), and his pitches had good speed diversity and great movement. The hitters were seeing a four-seem fastball leave Z's hand, but found themselves swinging at sinkers.

The Next Steps Pitch F/X
The sabermetric community has been really excited lately about the prospect of Field F/X (and Hit F/X, which is essentially a part of Field F/X). Field F/X -- sometimes called Fieldf/x -- would use the same (or similar) technology to Pitch F/X, but instead of capturing the movement of pitches, the cameras would observe how far and fast fielders get to and play balls. This would essentially rewrite the books on defense.

Recent reports have suggested Field F/X can and will do many amazing tings:
“We had shown them data capture on one play, a steal, and they kept asking us questions,” he said. “‘Do you get the initial lead? Do you get the secondary lead? Do you get the windup time, the pitch time, the pop time, the time it takes to throw down to second base?’” The answer in every case was yes.
It appears Field F/X could be just a few seasons away, which is no doubt thrilling.

1 comment: