
Introducing a new way to predict pitches: Type-Zone Modeling 🧵 medium.com/@devorsetza/the-a…

Resident Dick Allen supporter, Northwestern Data Science ‘25

Introducing a new way to predict pitches: Type-Zone Modeling 🧵 medium.com/@devorsetza/the-a…

We can use this type-zone area to predict any pitch with 25% accuracy. Do this for an entire season, and we find the predictability of every pitcher. Want to know who MLB’s most predictable pitchers are? Or the teams with the most-predictable pitch selection? How about how hitters fare against predictable pitches? Check it all out here: medium.com/@devorsetza/the-a…

My type-zone model is trained on more data than just repertoire and count, predicting the correct type-zone combo about 11% of the time with one circle. But we want to find back-up plans, alternate scenarios with a high likelihood of becoming real. Thus, we place two more circles that are trained to maximize probability, either by overlapping with our primary type-zone circle or by creating one for a different pitch type.

We can combine these models to find the most likely pitch type + pitch location combination in any given scenario. I’ll be calling this “type-zone combo”. With only pitch repertoire and count, you will predict type-zone combo 8% of the time. Say the most likely type-zone combo is a curveball down and away. This is great for knowing the most likely take scenario, but we’d like to account for mix-ups. We’d like to find potential target pitches for swing scenarios as well.

All you need to predict pitch type with good accuracy is the pitcher’s repertoire and the count. With that, you will correctly predict pitch type 39% of the time. The same inputs will correctly predict pitch location within a 6-inch radius 17% of the time. However, these factors are incomplete in isolation.

Lowest OPS by a ROTY-winning first baseman: Eric Karros, .730 Chris Chambliss, .749 Sal Stewart, .787 Karros was the only NL rookie to qualify for the batting title in 1992 Chambliss was one of two AL rookies with 400 ABs in 1971 Sal Stewart’s ROTY is a major outlier
How to lie with stats

Shohei Ohtani was worth 13.5 era-adjusted bWAR in 2022
Version 3.2 of our Era-Adjusted Baseball Stats is now live! eckeraadjustment.web.illinoi… Some of the best baseball in history is being played right now. Take a look! Version notes: - Stats are now up-to-date through 2026. - Talent pool version 3.0 remains the latest version. - The fullhouse R package will be updated to include the new stats. - Digressions document version 3.1 is still the latest, but its stats are outdated. I may update it eventually, but don't hold your breath. @PaulHembo @BaseballWRLD_ @JessicaDBrand @AlexDevMLB @GoBirdinals

Every outfielder with >45.0 WAR, an MVP, and a World Series ring HOFers: Babe Ruth Willie Mays Hank Aaron Stan Musial Tris Speaker Mickey Mantle Frank Robinson Rickey Henderson Roberto Clemente Joe DiMaggio Reggie Jackson Willie Stargell Joe Medwick Non-HOFers: Pete Rose Mookie Betts Christian Yelich Really makes you think
Traded in 2018 2018 NL MVP Captain 2026 World Series champion

Hunter Goodman has accumulated 4.5 WAR in his 4-year MLB career (per FanGraphs) This makes him the most valuable catcher in Rockies history, surpassing Chris Iannetta’s 4.2 WAR
Hunter Goodman is on pace for 41 home runs this season. Season Stats: 39 HR | 82 RBI .245 AVG | .312 OBP | .840 OPS Pace: 41 HR | 86 RBI #Rockies

2026 is the second season in 135 years were 7+ of the top-11 players in WAR are pitchers or two-way players (per FanGraphs)
Removing Ohtani, we have 6 pitchers in the top 11 in fWAR I can't remember the last time we had more pitchers than nonpitchers in the top 11 Indeed, I will just say, without any research at all: that has never happened!

Primary position by the top-30 hitters in fWAR C: 2 1B: 3 2B: 6 3B: 5 SS: 4 LF: 2 CF: 3 RF: 3 UT: 2 The last full season with more than 3 full-time centerfielders in the top 30 was 2015
War overrates center field defense so much that only one of them has even cracked 4.5

*I forgot the 2021 Braves. 99 wRC

*and the 2021 Braves, who had a 99 wRC+

The only World Series champions in the last decade without a 110 wRC+ was the 2019 Nationals They had a 104 wRC+, led MLB in starting pitcher WAR, and were top-5 in both position player WAR and pitcher WAR
Who is the one without a 110? And what was the #?

8 of the last 9 World Series champions had a team wRC+ over 110 The only team in 2026 with a 110 wRC+ of higher is the Chicago Cubs
I do not believe a sub-100 wRC+ team has won a World Series in the last decade

Cal Raleigh and Taylor Ward have seen their ISOs drop over 150 points from 2025 to 2026 Both are among the 20 biggest power decreases in MLB history