OpenCommand

Open-source command reports!

[OpenCommand @open_command]

I'll be using this account to publicly share the results of OpenCommand, a computer-vision based command tracker.

Ideas in mind:
- MLB command leaderboards
- Target maps
- Pitcher command trends
- Pitcher tactics changes
- Reply to this tweet if you have ideas!

[DT K @tomdoyo]

Some news:

I decided to open source my command project. It's called OpenCommand.

This means anyone can see the code (and even the data)!

I hope this gives more people a chance to pursue and share cutting edge baseball research.

Anyways, here's how to measure command: 🧵

1/N

[OpenCommand @open_command]

Thanks for reading!

Next thread will explain how OpenCommand becomes the biggest edge in fantasy since Stuff+.

Meanwhile, go follow @lukeblomm, the man behind!

[OpenCommand @open_command]

🚨 OpenCommand is updated!

This will be the LAST major update as OpenCommand has genuinely come extremely close to ground truth.

Honestly, this is the biggest edge in fantasy since Stuff+. The results are remarkable.

1/N

[OpenCommand @open_command]

Glove dependence ranges mostly between 0 to 1.

Only few pitchers are estimated to have glove dependence >1. I think part of this is due to smaller values functioning as regularization for glove detection noise.

6/N

[OpenCommand @open_command]

@SheistyJudge @yakyucosmo @MLB When he didn't hit the circle?

[OpenCommand @open_command]

Yoshinobu Yamamoto had the BEST COMMAND among starters so far in 2026.

On May 18th, his splitter averaged 5.19in miss, which is less than HALF of league average splitter (11.00in)!

Full leaderboard ↓↓

[OpenCommand @open_command]

@TOutmanSzn Sounds like somebody has the best command in baseball 👀

[OpenCommand @open_command]

@RiCatching @tomdoyo @MLB The targets are adjusted based on typical setup for each pitcher so I think that should be taken care of

[DT K @tomdoyo]

We can adjust for tendencies by offsetting the target by the avg pitcher × pitch type miss pattern.

E.g. Logan Webb's changeup averages 13in below the glove

This is the current math behind "inferred" targets.

But there are 2 huge flaws with it. Can anyone guess what they are?

10/N

[OpenCommand @open_command]

@ChrisBerejik @MLB Any ideas how?

[OpenCommand @open_command]

@OSCannons @MLB Correct, but we can get pretty close :)

[OpenCommand @open_command]

151. 10.24in MIA Eury Pérez
152. 10.25in TEX Kumar Rocker
153. 10.30in NYY Max Fried
154. 10.30in TB Shane McClanahan
155. 10.32in PIT Jared Jones
156. 10.32in LAA Walbert Ureña
157. 10.34in NYM Nolan McLean
158. 10.35in TEX MacKenzie Gore
159. 10.38in MIL Brandon Sproat
160. 10.39in ATL JR Ritchie
161. 10.41in HOU Hunter Brown
162. 10.49in PIT Bubba Chandler
163. 10.53in KC Luinder Avila
164. 10.53in NYM Freddy Peralta
165. 10.54in HOU Tatsuya Imai
166. 10.57in STL Andre Pallante
167. 10.64in COL Ryan Feltner
168. 10.66in WSH Andrew Alvarez
169. 10.70in MIN Connor Prielipp
170. 10.72in HOU Lance McCullers Jr.
171. 10.72in MIA Max Meyer
172. 10.77in TEX Jack Leiter
173. 10.78in LAD Shohei Ohtani
174. 10.81in NYM Kodai Senga
175. 10.86in AZ Mitch Bratt
176. 10.87in SD Lucas Giolito
177. 10.94in LAA Jack Kochanowicz
178. 11.04in MIN Taj Bradley
179. 11.19in LAA Yusei Kikuchi
180. 11.21in KC Cole Ragans
181. 11.24in CLE Joey Cantillo
182. 11.35in TOR Trey Yesavage

[OpenCommand @open_command]

101. 9.73in DET Troy Melton
102. 9.75in BAL Chris Bassitt
103. 9.77in KC Michael Wacha
104. 9.78in WSH Zack Littell
105. 9.78in NYY Cam Schlittler
106. 9.78in KC Noah Cameron
107. 9.78in WSH Jake Irvin
108. 9.79in STL Dustin May
109. 9.80in ATL Spencer Strider
110. 9.80in LAA Sam Aldegheri
111. 9.80in CLE Slade Cecconi
112. 9.82in BAL Trey Gibson
113. 9.84in MIA Ryan Gusto
114. 9.85in MIL Kyle Harrison
115. 9.85in CHC Edward Cabrera
116. 9.86in LAA Grayson Rodriguez
117. 9.86in SF Robbie Ray
118. 9.89in CWS Sean Burke
119. 9.90in LAD Emmet Sheehan
120. 9.91in MIL Shane Drohan
121. 9.92in TOR Dylan Cease
122. 9.92in CHC Matthew Boyd
123. 9.96in WSH Cade Cavalli
124. 9.97in MIN Zebby Matthews
125. 9.97in NYY Ryan Weathers
126. 9.97in CWS Davis Martin
127. 9.98in BOS Patrick Sandoval
128. 9.99in SEA Logan Gilbert
129. 9.99in LAD Tyler Glasnow
130. 10.00in PHI Andrew Painter
131. 10.01in LAA Reid Detmers
132. 10.01in BAL Kyle Bradish
133. 10.02in BAL Shane Baz
134. 10.03in LAD Roki Sasaki
135. 10.06in COL Gabriel Hughes
136. 10.09in COL Michael Lorenzen
137. 10.09in SF Trevor McDonald
138. 10.10in TOR Simeon Woods Richardson
139. 10.11in CIN Chase Burns
140. 10.13in BAL Dean Kremer
141. 10.13in ATH Gage Jump
142. 10.16in SF Carson Whisenhunt
143. 10.17in CWS Noah Schultz
144. 10.19in SD Germán Márquez
145. 10.19in DET Framber Valdez
146. 10.20in NYY Carlos Rodón
147. 10.21in KC Kris Bubic
148. 10.21in STL Hunter Dobbins
149. 10.23in BOS Connelly Early
150. 10.23in HOU Spencer Arrighetti

[OpenCommand @open_command]

@chefceej 🫡

[OpenCommand @open_command]

🚨OpenCommand is updated with new target detections!

This solves a major case where catchers use "decoy targets" with a runner on 2nd.

Turns out the solution was really simple:

1/3

[OpenCommand @open_command]

@Randolfson Thank you so much! I'm a big fan of KBO and am planning on making this work for KBO in the future. Stay posted 🫡

[OpenCommand @open_command]

I'll be using this account to publicly share the results of OpenCommand, a computer-vision based command tracker.

Ideas in mind:
- MLB command leaderboards
- Target maps
- Pitcher command trends
- Pitcher tactics changes
- Reply to this tweet if you have ideas!

[DT K @tomdoyo]

Some news:

I decided to open source my command project. It's called OpenCommand.

This means anyone can see the code (and even the data)!

I hope this gives more people a chance to pursue and share cutting edge baseball research.

Anyways, here's how to measure command: 🧵

1/N

[OpenCommand @open_command]

@PeachTree16 @TewksHitting Yes

[DT K @tomdoyo]

We can adjust for tendencies by offsetting the target by the avg pitcher × pitch type miss pattern.

E.g. Logan Webb's changeup averages 13in below the glove

This is the current math behind "inferred" targets.

But there are 2 huge flaws with it. Can anyone guess what they are?

10/N