Category Archives: PITCHf/x

THT: The Lurking Error in Statcast Pitch Data

Today at The Hardball Times, I wrote about pitch-tracking differences between PITCHf/x and Statcast. I found that Statcast’s radar is worse at tracking vertical movement and vertical location, but has similar ability for horizontal movement and horizontal location. And velocities are more accurate since Trackman’s out-of-the-hand readings replaced PITCHf/x’s 50-foot recordings. In addition to the general takeaways, there are interesting differences from park to park. You can read more here.

THT: Ground balls!

I published two articles on ground balls last week at The Hardball Times. The first used boosted trees to assess the impact that pitch and contextual factors have on predicting whether grounders will be hit. The second took a closer look at how pitching inside hardly scares batters into hitting more grounders, even though the Pirates are strong believers and have been a great GB team in the past three full seasons.

Bryan Morris’ Shift

Since his June 2014 trade to the Marlins, Bryan Morris has posted a 0.91 ERA with a 3.06 FIP in 49.1 IP.  Pretty good.  Obviously, you can anticipate that his numbers come back to earth–by Steamer’s rest-of-season projection, he’s expected to have a far more ordinary 3.46 FIP and 3.55 ERA this year–but Morris has made some changes to his mound positioning and approach that are worth a look.

Morris has shifted his position on the rubber, moving from the first base side to the extreme third base side.   This is shown in the two photos below; on the left, Morris pitches for the Pirates at PNC Park on 5/21/14.  On the right, he’s in his set as a Marlin in a 7/9/14 game at Chase Field.  I circled (what appears to be) the rubber on each mound.  Neither the pictures from the TV feeds nor the circles match up perfectly, but the shift is still plainly obvious.


That will change the point at which PITCHf/x picks up his release; the Brooks Baseball chart below shows Morris’ horizontal release point by game from the start of the 2014 season and onward.


By PITCHf/x, we can see that Morris’ horizontal release point moved over a foot closer to the third base side.  The first big drop came while Morris was still with the Pirates, so his experiment likely began before the trade.

Along with the change in where he toes the rubber, Morris is attacking hitters differently.  The two charts below show Morris’ pitch type locations against left-handed hitters (in the left image) and right-handed hitters (in the right image) during his time with the Pirates in 2014.


By PITCHf/x’s classification algorithm, Morris threw a lot of four-seamers inside and outside to both LHH and RHH.  He also featured a sinker with a slight concentration to the arm side, and sliders and cutters that often ended up catching plenty of the plate.  We’re left with a rather amorphous cloud of pitch locations.

We see a clear contrast in the charts below, which show all of Morris’ pitches in his tenure with Miami.  Once again, LHH are displayed on the left, and RHH on the right.


Against LHH, Morris has abandoned the four-seamer, going just about exclusively to the sinker running away to the arm side and the cutter breaking in toward the hitter.  He’s done a better job staying away from the heart of the plate, keeping pitches on the edges and out of the zone.  When facing RHH, he’s thrown way more cutters, again staying towards the edges and looking to induce chases.  He’s also kept his sinker low and/or inside, and occasionally mixed in four-seamers to the glove side.

(The last two sets of charts are via Baseball Savant.)