This is awesome! I've felt that Stuff+ models are less helpful than I'd hope because they are, literally, reductive. There's no new information presented to me if I find out Corbin Burnes' hard cutter grades out well by models. But by having an "open source" model, we can piece together the "why" more than the "what."
I find sinkers to be interesting in general because pitch classification is ultimately "made up," and sinkers (and cutters*) are the poster children for this. Ex: Josh Hader's sinker essentially isn't one - it's a four-seam fastball by another name and grip. It has ~20 inches of vert and ~7 inches of run, and he frequently throws it at the top of the zone with no other secondary fastball. Félix Bautista throws his fastball with a two-seam grip as well, it's just that it also has similar shape to Hader's (with a dramatically different release point) and he pounds the top of the zone as well. Interestingly, Brooks Baseball has him throwing both a 4SF and a sinker, but they are heavily clustered and separated by a few inches of horizontal break. The reverse of this is Paul Skenes' splinker, which is obviously being thrown as a fastball and not a true offspeed offering (Skenes himself apparently calls it a sinker).
*StuffPro/PitchPro made the leap and finally broke down cutters between hard cutters (Burnes, Clase, Jansen) and breaking ball cutters (nearly everybody else). I would be curious to see a SMOKE model for cutters, as well as changeups, which generally do not grade well by Stuff+.
Side note, what exactly is an arm-side pitch with gyro spin? A screwball?
It appears to work better against opposite handed batters, but sinkers are more effective against same-handed batters, and for that matchup more movement is better. So if you have a rising sinker, it needs to be spin efficient/move a lot, or you want a sinker with lots of sink, and then mostly throw it to same-side batters.
This is awesome! I've felt that Stuff+ models are less helpful than I'd hope because they are, literally, reductive. There's no new information presented to me if I find out Corbin Burnes' hard cutter grades out well by models. But by having an "open source" model, we can piece together the "why" more than the "what."
I find sinkers to be interesting in general because pitch classification is ultimately "made up," and sinkers (and cutters*) are the poster children for this. Ex: Josh Hader's sinker essentially isn't one - it's a four-seam fastball by another name and grip. It has ~20 inches of vert and ~7 inches of run, and he frequently throws it at the top of the zone with no other secondary fastball. Félix Bautista throws his fastball with a two-seam grip as well, it's just that it also has similar shape to Hader's (with a dramatically different release point) and he pounds the top of the zone as well. Interestingly, Brooks Baseball has him throwing both a 4SF and a sinker, but they are heavily clustered and separated by a few inches of horizontal break. The reverse of this is Paul Skenes' splinker, which is obviously being thrown as a fastball and not a true offspeed offering (Skenes himself apparently calls it a sinker).
*StuffPro/PitchPro made the leap and finally broke down cutters between hard cutters (Burnes, Clase, Jansen) and breaking ball cutters (nearly everybody else). I would be curious to see a SMOKE model for cutters, as well as changeups, which generally do not grade well by Stuff+.
Side note, what exactly is an arm-side pitch with gyro spin? A screwball?
Thank you for the kind words and for taking the time to read through.
Any pitch can have gyro spin, even a "fastball", aka the Justin Steele fastball.
Why wouldn’t you want to throw pitches with a low Spin Efficiency / Total Movement if it has better stuff +?
It appears to work better against opposite handed batters, but sinkers are more effective against same-handed batters, and for that matchup more movement is better. So if you have a rising sinker, it needs to be spin efficient/move a lot, or you want a sinker with lots of sink, and then mostly throw it to same-side batters.
Is mov_tot_nathan x mvt + z mvt?
mov_z_nathan is z mvt?
mov_z_nathan = 0.5 * [Atz] * SQUARE([tf]) * 12 from the spreadsheet.
mov_tot is the SQRT(Z²+X²)