Alex Caravan

MLB Data Science Manager @ Penn Interactive | ex-@DrivelineBB | (Small sample ⟦size⟧)

[Eno Sarris @enosarris]

Ten pitchers that have changed their arsenals the most https://t.co/SNzdmDYrTL

[Kostya Medvedovsky @kmedved]

👀👀👀

[Jack Lambert @jacklambert__]

MLB walk rate is 10% right now... up 1.6% from last year (obviously way too early, but if the season ended today, this would be 3rd highest all time!)

Is this related to ABS? Here's how AAA walk rate has changed since ABS was implemented in 2022

[Ben Resnic @BenResnic_]

Hitting fly balls further than expected is actually a repeatable skill. A ton of variance goes into how far a batted ball goes (wind, spin, elevation, etc), so the posterior is wide. In 2025, Javier Sanoja led all of MLB averaging 10.6 feet further than expected on his fly balls.

[Chris Langin @LanginTots13]

WBCでの太田泰示のリリース特性に感銘を受けて、日本人投手分析の2本目のブログを書きました。

リリース高130cm・エクステンション238cm・球速154 km/h

体感球速158.5 km/h。同バケット平均は146.5 km/h。

全文はこちら→ https://t.co/Lw9TQrFZSt

[Carson Evans @carsonevanss]

As we head into the 2026 season, let's take a look at some of the players who played in MLB in 2025 that have seen the biggest jump in their max EV so far in Spring Training.

🧵⬇️...

[FanGraphs Baseball @fangraphs]

InstaGraphs: RosterResource Now Has More Detailed Contract Info! https://t.co/AcPkgu8leq

[Tangotiger 🍁 @tangotiger]

#SABRanalytics

Great presentation by the Statcast team who I am lucky enough to work with on a daily basis ( @dohyoungpark , @_dadler , Graham Goldbeck)

[Alex Caravan @Alex_Caravan]

@enosarris I get into town tonight, I owe you a beer and a _ from chicago hahaha

[Brandon Dim @BrandonDim24]

🚨 ABS Impact Analysis: Shadow Zone Edition

Mike Trout gets 17.3% of in-zone shadow pitches incorrectly called balls - the most in MLB. With ABS coming, that advantage disappears.

316 hitters analyzed. Average 8% incorrect ball calls on shadow zone strikes. The transition will be significant.

📊 Full breakdown 👇
(Graphs on individual posts for better visibility)
*Disclaimer* ABS is based off of height, this analysis was based on "rule-book" zone from 2025

[Conner Pelletier @connerpelletier]

Last year I set out to see what biomechanics patterns are present in pitchers with good command and presented my findings @sabr. This project fundamentally changed the way I view command and repeatability.

The long overdue blog is out! Check it out here: https://t.co/Tcsp6w8CXf

[FanGraphs Baseball @fangraphs]

Introducing the FanGraphs Lab https://t.co/9iZtrnQuwy

[Mark Carey @MarkCarey93]

High blocks, low blocks, everywhere you look...blocks.

Using data from SkillCorner, I break down how each side likes to set up out of possession, and what it means for their playing style.

@TheAthleticFC

https://t.co/e3wF0StvGx

[Tej Seth @tejfbanalytics]

lowest pressure rate on true pass sets allowed over the past two seasons (min. 300 non spike pass block snaps)

[Eno Sarris @enosarris]

From 2021-2023, the league threw 10k more sweepers every year. Then batters began to make better contact & swing decisions against the pitch, because they were trained to see that movement. So we've refreshed Stuff+, taking 2020 and 2021 out of the training data. Sweepers down:

[Ted Knutson @mixedknuts]

Just to follow up on this conversation, Understat is probably the WORST of the xG models available. (Or whatever @xgphilosphy uses). It's just public and free.

This is an excellent analysis comparing different models/types of models.

https://t.co/KNpNGoEpUl

[Lau Sze Yui @903124S]

By conditioning on 4-seam with highest spin efficiency for each degree change of arm angle there will be a 0.9 degree change of spin axis. For typical 40 degree slot non-gyro FF will differ 6 degrees and for average FF it will differ by 16 degrees

[Jeremias Engelmann @JerryEngelmann]

In the more technical part 2 of our series on APM I share state-of-the-art tips and tricks -- which I've learned while working for two NBA teams -- on how to

- implement priors
- better handle penalization
- easily compute standard errors

https://t.co/UsJ22EI4DX

[Mays Copeland 💥 @MaysCopeland]

I've updated https://t.co/9Hqtel8sNm for 2025, if you want to see how your favorite projections did.

I'm planning to write-up a brief summary for those who just want to know which projections to use next year.

[Lau Sze Yui @903124S]

By training an out probability model we can plot distribution and sun on average ~2% to out probability. 8/n