Trusting The Robots
Finding Early Values & Risks Using Stuff Models
Ah, STUFF models. What is one to do with them? Do they actually matter for fantasy?
Jackson Jobe of my Tigers is a fantastic example of why so many great fantasy players make jokes like, ok wow a 245 fastball score but I don’t play in a Stuff+ league. Or ok, what does the other model say and is it scaled the same? Or, have the inputs changed this year? Jobe lights up models but has really struggled to strike anyone out. Just use K-BB% to decide if a pitcher is good, the doubters say. Or how about swinging strike overall and for individual pitches. If the model loves a pitch but it doesn’t miss bats, did a tree fall in the woods? You get the idea. There’s more to succeeding as a major league pitcher than crafting a shape.
On the other hand, here are the top 7 starters by FanGraphs’ Bot Stuff as the week began: Skubal, Framber, Sanchez, Greene, Peralta, Crochet, Glasnow. Pretty good list.
One of the stuff models’ more vocal critics, Jeff Zimmerman of FanGraphs, decided to find out if they had any utility for fantasy (or really, predicting performance generally). I won’t spoil his work or income by revealing many specific values (the full results are in the new appendices of The Process book, link below)—but in short, yes, the models can tell us some useful things. Despite no adjustments since it was introduced, the FanGraphs Bot performed best among a series of other models and did well in correlating to various statistics when compared to K-BB% or SIERA, especially in small samples. For example, Jeff found that with one month of data the Bot was best at predicting ERA, K% and HR/9 for the rest of the season. The Pitching+ model also did well.
If you’re using projections, the OOPSY system incorporates Pitching+.
So, with that background and the caveat that we are still a few starts or appearances away from one month of data, let’s take a look at a few pitchers who stand out for good or ill.
***
Luis L. Ortiz - Stuff UP
More like Luis “W” Ortiz, amirite? The Guardians offseason acquisition has gross surface numbers thanks to a disastrous first start at San Diego, but he’s been much better since, including a 1 hit, 2 walks, 1 earned run outing with 10 strikeouts over 5.2 innings against the Royals. The Ks caught my attention (and faab bids) as KC is one of the more contact-oriented offenses.
Ortiz is showing greater velocity on both his fastballs (sitting over 95) along with greater arm side run and drop on his changeup. All three are performing better per the models with his overall Stuff+ rising from 94 to 102 and BotStf from 54 to 57 (the Bot is on a 20-80 scouting scale; 50 is average). From Zimmerman’s research this equates to a 3.73 ERA, 1.18 WHIP and 18% K-BB.
Cleveland is having Ortiz throw his four seamer almost 9% more often while reducing the sinker usage about 11%. That’s a recipe for more whiffs. The changeup usage is up too (8.9% from 0.9%) and while its swing and miss is modest (18%), it has yet to allow a hit and batted balls in play have a -10 launch angle. This gives Ortiz a true weapon to neutralize lefties; previously he really only had a cutter, which was hit for decent damage last season (.442 xSLG).
Ortiz has always thrown pretty hard but the whiffs have been lacking. If he can keep the velo gains he’s shown while throwing more four seamers we could get a really nice season from him. My fantasy rotations will be happily rounded out with a sub-4 ERA, sub-1.20 WHIP and 26% K rate with potential for further growth.
Jose Soriano - Stuff DOWN (But…)
The Angels Soriano has traded about a mile an hour of fastball and splitter velocity for greater control, although in the early going his walk rate is up slightly (9.6 to 10.7%). That is obviously high but he has also raised the strikeouts by almost 4% (24.3%). On the surface this is the best he’s ever been with a 2.70 ERA and 1.00 WHIP.
This is true despite the fact that the models don’t like Soriano’s stuff as much. That makes sense as velocity is a key component to these models but in this case he may have had enough to spare—i.e., his sinker still sits 96.9, 90th percentile overall and third overall among starters. He has throttled down the slider almost 4 mph, which may be a positive in spatial separation even if models don’t love the change on its own.
Specifically, Soriano’s Stuff+ is down from 109 to 95 and the Bot docks him from 59 to 53. But his Location+ is up 13 points and Bot Command went from 38 to 44. These are probably good if not necessary adjustments, and they’re paying off in other ways: the slider is getting 10 more inches of drop, the curve 4” more. The whiffs are way up on the slider (58%) and the splitter (50%).
His fastballs are moving less horizontally, which tells me he’s able to command them better. The sinker especially seems tighter but it’s very early to glean much from heat maps. Let’s do it anyway!
2024:
2025:
Notably Soriano’s ground ball rate, already elite, is up to an incredible 67%. We need more sample there but it was 60% last year. That will always be a strength.
Now yes, the stuff numbers are down but based on the Bot correlations, Soriano’s already projecting to a 3.90 ERA and 1.22 WHIP with a 24% K rate. That’s pretty good for his cost, which was almost nothing. I also suspect the ERA value may be a bit too high for a power sinker baller like this putting over 6/10 balls on the ground. Some concomitant WHIP risk makes sense.
The key to a full breakout will be Soriano’s walk rates correlating better to his stuff—based on The Process research, he should settle closer to 8% BB rather than the 10% he has presently.
Bryce Miller - Stuff DOWN
Just a quickie on Miller, who I wrote about in the last post. But he really fits this theme and has gotten off to a less than inspired 4.50 ERA and 1.44 WHIP. He’s limiting damage (.56 HR/9, 2.1% BRL), but is not striking guys out (19%) or corralling walks (11.8%). The surface numbers would be one thing but in his endless quest to tinker more than any other starter in history, I fear he’s gotten away from what made him really good in the first place. When you watch him it sure feels like the plan is muddled.
The Bot thinks Miller’s sinker, splitter and knuckle curve are all worse (Stuff+ agrees except that the sinker is unchanged). Once able to punch guys out with his four seamer at a high rate (13.5% SwK each of first two seasons), it’s returned a more average 9.8% so far. The only pitch missing bats is the curve (27% SwK), and that at an elite rate. It’s also been hit around (.385 BA), a strange combination. Or maybe not as it’s dropped from a 50 grade pitch to a 43, per the Bot.
Miller should be better and mostly solid? His draft price was just a little high. Notably he’s helped by the Mariners park more than any Seattle starter (1.96 ERA home, 4.07 away last season). Indeed this season he’s allowed 4 earned in 10.2 home innings and 4 in 5.1 IP away.
At minimum right now I’m picking my spots in road starts until we see more promising underlying signs, draft cost be damned.
Davis Martin - Stuff SAME(?) But…
I know, you’re thinking DAVIS MARTIN? This double-first named middler isn’t even on my radar.
I have a confession. I own Martin on 3 teams. Yes, he’s my last SP by priority or at least how I conceive of their values, but I think there’s something here.
Martin’s BotStf is 52 from 53 last year, so overall about the same. But note while everyone’s gaga for Shane Smith, it’s Martin with the best stuff grade of the ChiSox starters. He now has a decent sinker (54 BotStf) and Martin was one of the first adopters of the kick changeup. That pitch has recorded a 62 BotStf, which ranks top 10 in MLB. (Smith has a good cambio too but it ranks 16th at 59).
Martin’s change has performed well with a 16% SwK. The problem is none of his other pitches are missing bats (FF at 8% is next best). Especially his slider, which returned a 38% whiff and .117 BAA last season but now has 6.3% and .343 marks, respectively. He is getting almost 4” more horizontal break on it but based on heat maps it appears he’s yanking it (easy takes) or it’s sitting dead zone. I think he can get it back? The pitch was also good in his limited 2022 sample (41% whiff).
I’m holding Martin in deeper leagues hoping he puts together the better walk rate (down 4 points to 6.8%), keeps this changeup rolling, and rediscovers the slider. But not tightly. If a better option comes along and we don’t see progress, adios.
Mason Montgomery - Stuff UP
This seems incredible and incendiary to even propose, but it’s possible top closer Mason Miller of the “ATH’s” isn’t even the best “Mason M” bullpen arm in baseball. Montgomery’s 79 Bot Stuff is matched only by Griffin Jax and exceeded only by Aroldis Chapman (still throwing 80 grade filth). But Montgomery (72) has much better command than Aroldis (51). The lesser known Mason M shares with Jax the honor of the only pitchers in baseball popping 80 scores in Bot Overall.
In addition, by the more famous Stuff+ Montgomery stands alone at 133 and second in Pitching+ at 144. Jax’s stuff component doesn’t come close there but his command score is so good he’s at a 147 Pitching+. Whatever weight one wants to assign these models, the fact Montgomery rates with that company is impressive.
Jax gets there with a varied starter-level pitch mix and even Chapman uses four pitches. Montgomery has two: the highest graded four seam (77) and a top 10 slider (68). With 70 grade command. It’s bonkers.
That control boost would be huge if Montgomery can maintain it. Last season he walked 5 in 9.2 IP (13.5%) and registered a solid but mortal 58 BotCmd. So far in 5 IP he’s walked none. If that 14 point jump is real (Location+ also up from 102 to 111), we’re looking at a real terminator.
As is, since debuting last season Monty has a 2.72 ERA, 1.15 WHIP and 14.9 K/9. He did get tagged for homers in back to back games in early April, which ballooned his current ERA to 4.50 (2.15 xFIP). He did show some vulnerability to the long ball in the minors, though that was mostly as a starter. What does that 79 BotStf model value tell us? Well, Zimmerman’s chart ends with a BotStf of 67 both because it is more starter-focused and well, very few guys exceed that rating. Based on that research, a 67 corresponds to .9 HR/9—hope that Montgomery can rein in the dingers.
We are all aware of Pete Fairbanks’ health risks. Perhaps relatedly, despite an early 1.50 ERA (3.36 xFIP), Pete’s stuff grades have declined since 2022 along with his velocity, which is now a full 2 mph lower than it was three years ago. His walk rate is up to 12%.
This is a long way of saying Montgomery is looming as a potential closer replacement in Tampa. They have other good arms like Edwin Uceta, Hunter Bigge and Manny Rodriguez. Manny in particular has been excellent. But Montgomery is on another level, at least according to the robots. I have added him to a couple teams for the ratios, strikeouts and potential he ends up closing games.
Reminder: we’re still a couple or three starts and appearances away from these numbers reaching a more reliable confidence value. But it’s definitely worth tracking for breakouts as well as disappointments (see below).
***
Other Bot Stuff Standout Names (on April 14)
Starters
Freddy Peralta 54 ⬆️ 65 🔥
Cris Sanchez 59 ⬆️ 64
Hunter Greene 54 ⬆️ 64 🚀
Mackenzie Gore 62 ⬆️ 63
Garrett Crochet 70 ⬇️ 62 (😬)
Dylan Cease 67 ⬇️ 62
Max Meyer 59 ⬆️ 61
Jacob deGrom 64 ⬇️ 60
Jack Leiter 58 ⬆️ 60
Kevin Gausman 50 ⬆️ 58
Best SP Bot Command:
Bryan Woo 68 (min. 10 IP)
Aaron Nola 68
Jameson Taillon 66 (👀)
Worst SP Bot Command:
Roki Sasaki 32 (🫠)
Corbin Burnes 36 (😬)
Andre Pallante 38 (???)
***
Bot Stuff Reliever Notables
Jose Alvarado 62 ⬆️ 78 🔥
Porter Hodge 70 ⬆️ 72
Robert Suarez 63 ⬆️ 68
Seth Halvorsen 65 ⬆️ 67
Jeremiah Estrada 68 ⬇️ 66
Orion Kerkering 68 ⬇️ 66
Tanner Scott 74 ⬇️ 65 (😬)
Randy Rodriguez 57 ⬆️ 65
—>
Here’s the link to Zimmerman & Bell’s The Process. You can get a pdf with all the new research for $14.95! (Jeff doesn’t even know who I am so this is just me loving the content):
https://www.thefantasybaseballprocess.com/