Pinned Tweet
An article digging into my Stuff+ and Pitching+ models, which can assess a pitcher's talent level from a small sample of pitches. Both are significantly more predictive than traditional stats, and perform better than public peers I was able to evaluate.
Freddy Peralta's stats pre & post trade are interesting.
Focus on results / ERA? He was a totally different pitcher in September.
Focus on xFIP, K-BB%? No better and he just got luckier.
Look at his Stuff+/Pitching+ model scores? Way better in September. Fastball velo and shape improved (beyond the IVB Trop impact) and changeup shape evolved a bit in a way the model likes.
Stating the obvious but it'll be fascinating to see what Paul Skenes looks like in terms of stuff next year. Should be able to get a sense in spring / early in the season if 2026 was a blip or his new normal. If 2H 2026 is his new normal, I suspect there is some physical issue, but either way I imagine that'll need to be resolved before a team will commit much in a trade with the Pirates.
Paul Skenes salary will be out of the Pirates comfort zone after next season and he will likely be playing somewhere else in 2028, per @JPerrotto
(pittsburghbaseballnow.com/pi…)
Jacob Young is an Anthony Volpe, probably a bit worse, bat - great defender but not sure what the Yankees would want with him. Question of course in 2027 will be if Judge will be playing RF regularly or they’ll need an additional regular OF beyond Jones, Belli, and Judge
Who would you go to in a key at bat? I know managerial decisions especially in the playoffs are a roll of the dice, but kind of wild. Same with the Blake Snell decision when his command and stuff was dropping off…
Seeing posts about the Yankees & Seiya Suzuki. Beyond defensive alignment, a concern is Suzuki had a good year probably propped up by luck while underlying power metrics declined. Avg exit velo fell ~3mph. Bat speed didn't, but this is a yellow flag in a guy turning 33 in '27.
I loved what we saw from Will Warren out of relief late in '26. His velo & FB shape improved significantly and so did outcomes. If he could keep this up, he'd be a truly elite leverage arm. If the Yankees rotation is fully healthy in '27, I'd like Warren in the bullpen. He shows signs of fatigue as a SP, with velocity & stuff falling off sharply after a few innings, and so a RP role seems to work well for him.
There was nothing about Jazz Chisholm's 2026 that suggests he got unlucky - if anything he got slightly lucky with wOBA ahead of xwOBA. I also see Jazz's 2025 as likely a career year where he outperformed his skills. But compared to Volpe, there is history suggesting Jazz can be a much better player than he was in 2026, with offensive output in the .320-.330 wOBA range (vs ~.290-.300ish for Volpe).
Now is that kind of output difference worth whatever Jazz would receive, even the QO of ~$23mm - very debatable. And this does not capture the "intangibles" or baseball IQ that perhaps Jazz lacks, or fans perceive he lacks anyway.
I root for Anthony Volpe, but the data is clear about the level of hitter he is, and it's tough to accept as an everyday player. In '26, the Yankees got so little out of 2B/3B that he wouldn't be a downgrade, but I think penciling in Volpe is giving up hope.
His skills have fluctuated a bit between contact and power, and he's made better swing decisions in 2025/2026 than prior years, but whether you look at wOBA, xwOBA, or my own metric Skill wOBA (attempts to capture offensive skill, cutting out the noise, and present it on wOBA scale), he's around a .290-.300 wOBA player each year. In 2026, in September he also showed a similar .300ish wOBA/xwOBA/Skill wOBA. There is no sign anything about his capabilities changed when he came back up late in 2026.
I root for Anthony Volpe, but the data is clear about the level of hitter he is, and it's tough to accept as an everyday player. In '26, the Yankees got so little out of 2B/3B that he wouldn't be a downgrade, but I think penciling in Volpe is giving up hope.
His skills have fluctuated a bit between contact and power, and he's made better swing decisions in 2025/2026 than prior years, but whether you look at wOBA, xwOBA, or my own metric Skill wOBA (attempts to capture offensive skill, cutting out the noise, and present it on wOBA scale), he's around a .290-.300 wOBA player each year. In 2026, in September he also showed a similar .300ish wOBA/xwOBA/Skill wOBA. There is no sign anything about his capabilities changed when he came back up late in 2026.
Jacob Misiorowski didn’t have his best stuff, especially fastball shape (IVB 1.6” lower than normal). Velo dipped in the later innings, perhaps in response to early command issues. Location score doesn’t look that bad overall, but he struggled to locate his slider.
For Spencer Jones to be an elite hitter, I think the clearest path is continued swing decision improvement. Jones has made steady progress in 2025-2026. His rate of laying off pitches he should take improved significantly from 2025-2026 in AAA. Where he is elite is attacking hittable pitches, where he made steady progress in AAA in 2026 and then further progress in the majors.
If he can take a step forward in 2027 in laying off pitches outside the zone and being more patient with marginal strikes early in the count, then he'll have well above average swing decision skills.
I've improved my Stuff+ & Pitching+ models. Performance is up slightly in most areas, with sizable gains for Pitching+ and on a pitch level, especially for rookies.
My Stuff+ is a blend of a swing leg & a location-blind distillation of Pitching+. Either leg usually has greater predictive value than public models I've tested, but the combination is additive. I'll post the swing vs distill breakdown, as well as Pitching+ and rookie pitch level performance, in the replies.
The new scores are live on the site (162iq.net). The live model is trained through June 28, 2026, while the benchmarking here is entirely on a version trained through 2025 (2023-2025).
The 162IQ Pitching+ model sees a bigger performance improvement in the update, and it outperforms peers by a significant margin. The model updates I made reduced noise on batting balls in the training process and incorporated learnings from my hitter model work (Skill wOBA), and this seems to have added more value when location is included.
My Pitching+ is a blend of a location-blind swing, location-blind distill, and a location-aware run value focused model. This blended approach where not all of the legs have location information significantly outperforms a simpler, fully location aware Pitching+ architecture.
Pretty crappy performance from Max Fried tonight. Lots of pitches hung over the middle of the plate, and his stuff just isn’t good enough to allow for that. Got lucky early on but it was only a matter of time. I would have gone Rodon and then Cole, but oh well.
Chris Sale’s strikeout stuff was electric early on, but the Dodgers elevated his pitch count quickly tonight and were able to hit two home runs in the 4th off of some solid pitches. The Freeman homer on an inside sinker is particularly impressive.
The Yankees defense is the key issue so far, and it’s pretty remarkable the Rays only scored 1 run in 2 innings. But while Schlittler has great stuff tonight, the Rays have just 2 whiffs (8% whiff rate) so far. Cam had a 40% whiff rate vs the Red Sox last week.
