Projection Fight Club 2024

BRYAN TERRY/THE OKLAHOMAN/USA TODAY NETWORK

The first rule of Projection Fight Club is that you don’t talk about Projection Fight Club. Fortunately, the second rule of Projection Fight Club is that you are allowed to write about Projection Fight Club — otherwise, I might get in hot water with our editors for pitching an article I can’t actually produce.

If you’ve been paying attention to the projections housed here at FanGraphs (this is an odd article to read if you’re haven’t been), you’ve probably seen that our player pages now include ZiPS, Steamer, and THE BAT projections for the 2024 season (ATC projections are also available). You may have compared them a little. Perhaps you’ve even shaken your fist at the heavens for the temerity of allowing these systems to besmirch the good name of your favorite player or team. For me, the most interesting projections are the ones where the various systems disagree the most. After all, we watch heavyweight fights, not heavyweight agreements. Nobody would shell out cash to watch the Universal Amiable Concordance Championship.

Since we now have the different projections available, I thought I’d highlight some of the players who inspire the greatest discord amongst the various systems. I’m not going to guess which system will end up being right — it would be inappropriate for me to write a piece like that with one of the pugilists in the ring — but where possible, I’ll talk a bit about the complications involved with projecting those players, and in the instances where ZiPS stands alone as the biggest outlier, I’ll try to lend some additional insight as to why my system is being so nice or mean.

It’s always a bit awkward measuring dispersion with three or four data points, but the idea here is to get a simple, rough ranking of where the projection systems disagree. For that purpose, I’m going to with standard deviation. For hitters, I’m using wRC+, while I’ll look to ERA for the pitchers, as we don’t have an ERA+ type number for pitchers on our pages. Again, the exact numbers aren’t super important to gauge disagreement. I’m going with 300 plate appearances on our depth charts for hitters and 250 batters faced for pitchers in order to ensure we get the most interesting disagreements, rather than a sprinkling of 27-year-old org guys at Double-A. Let’s start with the hitters:

Top Projection Disagreements – Hitters
Name ZiPS wRC+ Steamer wRC+ THE BAT wRC+ The BAT X wRC+ StdDev
Wyatt Langford 118.4 121.1 94.5 94.4 14.6
Michael Busch 112.2 97.9 88.6 86.3 11.8
Shohei Ohtani 138.0 144.7 162.7 159.4 11.8
Kyle Manzardo 112.8 112.7 95.4 95.3 10.0
Mark Vientos 99.4 108.2 87.0 88.8 9.9
Triston Casas 126.9 129.5 115.7 108.9 9.7
Matt Chapman 117.2 103.6 108.6 124.4 9.2
James Outman 113.8 102.9 99.1 91.8 9.2
Tony Kemp 93.5 97.3 85.4 77.5 8.8
Brett Baty 99.1 105.0 87.2 87.5 8.8
Nick Senzel 66.1 84.0 81.1 84.9 8.8
Mike Trout 136.5 133.1 148.4 150.9 8.7
Vidal Bruján 86.5 83.8 70.5 71.2 8.3
Isaac Paredes 123.6 127.3 127.1 109.9 8.2
Oscar Colás 85.0 91.1 73.5 75.9 8.2
Adam Duvall 103.9 87.2 96.3 87.0 8.1
Brandon Belt 121.9 106.0 109.4 104.3 7.9
Santiago Espinal 93.3 97.8 82.3 82.3 7.9
Abraham Toro 98.0 100.7 85.3 86.5 7.8
Tyler O’Neill 120.3 105.4 103.0 111.0 7.7
Steven Kwan 102.5 113.2 106.1 95.0 7.6
Nick Madrigal 93.8 98.1 86.2 81.4 7.5
Jeff McNeil 107.3 114.1 106.1 96.2 7.4
Avisaíl García 80.3 84.1 85.6 97.3 7.3
Jack Suwinski 117.8 107.3 102.6 102.0 7.3
Aaron Judge 159.8 157.4 171.9 170.3 7.3
Nolan Arenado 110.9 112.4 113.1 97.9 7.2
Will Benson 101.6 99.5 88.3 88.6 7.0
Colt Keith 103.4 105.9 92.6 92.5 7.0

In the case of Wyatt Langford, there’s no one outlier, with ZiPS and Steamer roughly agreeing on the positive side and THE BAT versions being quite bearish in the short-term. I try to avoid projecting players after one partial minor league season, but Langford is one of the exceptions, as he seems to be very much in the Rangers’ immediate plans. Thankfully, he’s a hitter, he’s coming out of college instead of high school, and he played in a major conference, making the projection less scary than it otherwise would be. Langford is one of four SEC hitters who ZiPS thinks has the potential to hit in the majors very quickly based on their college stats (Langford, Dylan Crews, Jac Caglianone, and Jace LaViolette, who has a great name for a detective in a Southern Gothic murder mystery). ZiPS is hedging a bit on Crews in the short-term since he didn’t exactly dominate in the minors in his cup of coffee and was really aggressive at the plate. Langford, on the other hand, looked extremely polished in his first stint. There’s probably more work to done in trying to project guys like this. It always proves tricky — you have a bunch of data that does a really poor job of predicting the future (amateur stats) and only a tiny sliver of the better stuff (minor league stats).

ZiPS is the only system excited to see Michael Busch in the majors in 2024. I’m guessing that the Cubs align more with ZiPS than the others here; otherwise, I don’t think they’d be as quick to carve out a role on the roster as they appear to be. It’s hard to say exactly why the other systems like Busch less than ZiPS, but even with a rather poor debut in the majors, my system gives him a combined translation/majors line of .251/.336/.477 for 2023. That’s higher than his projection, of course, since his translation line was .239/.310/.424 the year before.

ZiPS is also the grumpiest of the three systems about Shohei Ohtani. My theory here is that it may reflect how the system treats injuries. ZiPS doesn’t know the full extent of his elbow injury, but it does know that a recent oblique injury cost him a month of the season. If I tell ZiPS that he wasn’t injured at all, his projected wRC+ goes from 138 to 147, which aligns more closely with the forecast he gets from Steamer. But even when I remove the injury penalty, there’s still a gap between ZiPS and THE BAT, which might come down to how the two systems deal with exit velocity data, though you’d have to inquire with creator Derek Carty about that one.

Kyle Manzardo is another ZiPS/Steamer vs. THE BAT fight. It would be interesting to go back and see whether there was a consensus among the systems on him going into 2023. After a breakout 2022, Manzardo was less exciting last year, due in large part to a shoulder injury that cost him a huge chunk of the season. ZiPS knows about the injury but still likes him — based on the hit data it has access to, it believes that his BABIP was significantly lower than it ought to have been given his profile.

For Triston Casas, my best guess is that the systems have different ideas about how a left-handed power hitter will generally fare in Fenway Park, which has a rather unorthodox shape and is generally an awful place to be a lefty slugger. Even Ted Williams felt the effects!

Moving on to Matt Chapman, I suspect what we’re seeing here is a difference in how the systems use Statcast data. ZiPS thought Chapman got rather hosed when it came to his power last season. He hit three 400-foot outs in 2023, and tied with Bobby Witt Jr. for the second-most fly ball barrels that turned into outs with 16. What’s really scary about this stat is that Ronald Acuña Jr. led the league with 17, meaning he put up those video game numbers while perhaps being unlucky, at least by this measure! Both Witt and Acuña had over 100 plate appearances more than Chapman did.

Speaking of players whose hit data indicates they got hosed last year, ZiPS thinks that Tony Kemp was absolutely robbed on his line drives, which is the biggest reason it sees a bounce-back season in store for him. For a team that could use a spare outfielder who can also play second base, he might be a really good pickup on a low-key deal.

Next up are the pitchers who inspired the greatest projection disagreements:

Top Projection Disagreements – Pitchers
Name ZiPS ERA Steamer ERA THE BAT ERA Mean StdDev
Shōta Imanaga 3.55 3.83 5.52 4.3 1.07
Brock Stewart 3.19 3.45 5.13 3.9 1.05
Bryan Mata 5.60 4.75 4.08 4.8 0.76
Joe Kelly 4.46 3.13 3.18 3.6 0.76
Carlos Carrasco 5.62 4.40 5.70 5.2 0.73
Kody Funderburk 4.17 3.61 2.86 3.5 0.66
Erik Miller 4.79 4.17 3.51 4.2 0.64
Nick Martinez 4.04 4.67 5.21 4.6 0.59
Luke Weaver 5.33 4.23 5.12 4.9 0.58
Eric Lauer 4.40 4.97 5.52 5.0 0.56
Cal Quantrill 4.93 5.92 5.87 5.6 0.56
Kyle Muller 4.67 4.05 5.13 4.6 0.54
Jorge López 4.70 3.89 4.92 4.5 0.54
Tim Hill 5.45 4.44 4.65 4.8 0.53
Tanner Rainey 5.00 4.28 5.30 4.9 0.53
Patrick Corbin 5.45 4.88 5.92 5.4 0.52
Jalen Beeks 5.07 4.30 5.29 4.9 0.52
David Peterson 4.49 3.50 4.25 4.1 0.52
Drew Rucinski 4.79 4.88 5.70 5.1 0.50
Julian Merryweather 3.97 3.53 4.52 4.0 0.50
Robert Stephenson 3.23 3.94 4.15 3.8 0.48
Aroldis Chapman 3.91 2.96 3.48 3.5 0.48
Ryne Nelson 4.32 4.97 5.25 4.8 0.48
Mark Leiter Jr. 4.32 3.84 4.79 4.3 0.48
Chase Anderson 6.21 5.26 5.85 5.8 0.48
Jimmy Lambert 4.94 4.87 5.72 5.2 0.48
Andrew Bellatti 5.20 4.43 5.28 5.0 0.47
Gregory Soto 4.52 3.60 3.90 4.0 0.47
Sean Newcomb 4.86 4.00 4.12 4.3 0.47
Austin Voth 4.30 4.23 5.07 4.5 0.47

There appear to be differences in how the systems treat NPB players. While he just missed this list, there’s considerable disagreement on Yoshinobu Yamamoto, with THE BAT again taking the pessimistic side of the bet. I can’t say how the other systems use the numbers, but ZiPS has every player who went from NPB to the majors/minors or the other way around in its database; the same goes for the KBO and Cuba Serie Nacional, though there are fewer players to go on. ZiPS also uses data from DeltaGraphs, which is the best public resource for Statcast-like NPB info, and while it’s not quite as transparent as Statcast, beggars can’t be choosers. Without that data, ZiPS would project Shōta Imanaga for an ERA about two-tenths of a run higher in 2024. It would like him even more if he were a groundball pitcher. I had interesting discussions — sadly, not for attribution — with analysts from three teams about Imanaga’s pitch profile over the course of the winter, mainly out of personal curiosity. Two saw Imanaga the way ZiPS does, while one took the view THE BAT does. You’ll have to wait a couple years to see who was right!

THE BAT clearly does not trust Brock Stewart’s strikeout bounce, projecting a 13.4% strikeout rate compared to 21.1% for Steamer and 24.2% for ZiPS (he posted a 25.7% mark for 2023). This may come down to Stewart’s very short résumé. He missed significant time due to Tommy John surgery, and his missed years due to injury were probably better than his 2018 and 2019 seasons! ZiPS is aware of the small recent sample, so it leans heavily on plate discipline data, and whatever your feelings are on the sustainability of his 2023 performance, it’s hard to fake a nearly 20% swinging strike rate and a 60.6% contact rate. Just to contextualize these numbers, Stewart turned the average hitter into someone with considerably worse plate discipline than Javier Báez. Stewart’s quite an edge case given his history, so ZiPS could very well be guessing wrong.

The Bryan Mata disagreement is a simple one — ZiPS is still projecting him as much more of a starter than the other systems that are using the depth chart’s playing time projections. Carlos Carrasco faces pretty much the same issue, with Steamer projecting him more as a reliever than ZiPS or THE BAT. The divergence on Joe Kelly might be how ZiPS calculates aging (by comparing Kelly to other old relievers with similar baseline performances), plus ZiPS sees his strikeout rate in 2023 as a little out of sync with the plate discipline data. Even as a groundball pitcher, ZiPS is less hopeful than its compatriots that Kody Funderburk can avoid a significant correction in his home run rate (he only allowed one homer in 52 innings in the minors last year).

Patrick Corbin is one of the more interesting departures here. It would be extremely enlightening to dig into Steamer and THE BAT and contrast how they deal with ERA/FIP interactions, though I’m guessing I will never get to do so directly. Post-peak Corbin is quite an odd pitcher. He’s still putting up WAR numbers that aren’t totally terrible, but he’s doing so with an ERA considerably worse than his FIP (and especially his xFIP) over the last four years. There’s not really a projection beef among his homer, walk, and strikeout rates. Rather, the difference here is that ZiPS projects Corbin’s ERA to be 0.23 runs worse than his FIP, while THE BAT sees him as 0.49 runs worse, but Steamer is projecting an ERA 0.11 runs better. Nobody’s projecting him to be good, mind you.

Let’s finish this off with the players who generate the most agreement. No comments here, just for historical purposes! OK, maybe one short note: I’m quite surprised that there’s relative consensus surrounding Henry Davis, who has had a real rollercoaster of a career as a prospect.

Top Projection Trilateral Peace Agreements – Hitters
Name ZiPS wRC+ Steamer wRC+ THE BAT wRC+ The BAT X wRC+ StdDev
Anthony Rendon 114.1 114.2 112.9 115.0 0.9
Alex Bregman 128.9 130.1 128.1 129.8 0.9
Garrett Cooper 100.8 100.2 102.2 100.3 0.9
J.D. Davis 106.9 105.6 107.5 107.6 0.9
Ketel Marte 122.3 120.3 121.6 122.4 1.0
Carlos Correa 115.9 118.2 116.2 117.3 1.0
Edouard Julien 114.6 114.1 114.9 112.5 1.1
Patrick Bailey 81.7 82.9 80.6 80.7 1.1
Henry Davis 102.8 102.6 102.9 100.6 1.1
Bryan De La Cruz 100.1 99.2 99.8 101.8 1.1
Michael A. Taylor 83.1 84.3 85.2 82.6 1.2
Mike Yastrzemski 105.5 105.3 104.5 102.8 1.2
Brandon Marsh 95.1 98.0 97.0 96.8 1.2
Yoán Moncada 102.1 101.1 101.9 99.4 1.2
Yordan Alvarez 166.6 168.9 168.6 169.6 1.3
Matt McLain 108.6 108.6 106.0 107.1 1.3
Freddie Freeman 139.9 142.9 141.7 142.5 1.3
Alex Verdugo 106.3 106.7 103.8 106.6 1.4
Eduardo Escobar 85.8 84.0 83.1 82.8 1.4
Ryan Noda 104.4 107.7 105.9 106.0 1.4
Nick Castellanos 99.7 100.7 102.6 102.6 1.4
Tyler Stephenson 96.7 95.0 93.5 94.1 1.4
Robbie Grossman 95.5 94.4 93.3 96.6 1.4
Ji Man Choi 103.8 104.9 102.7 101.6 1.4
Jon Berti 88.3 91.1 88.1 88.5 1.4
Elly De La Cruz 92.3 95.8 94.7 94.5 1.5
Lourdes Gurriel Jr. 108.3 108.7 109.7 106.2 1.5
Riley Greene 113.0 115.6 112.5 112.3 1.5
Anthony Volpe 95.6 98.4 95.0 97.3 1.6
Gleyber Torres 120.9 121.6 118.1 120.9 1.6

Top Projection Trilateral Peace Agreements – Pitchers
Name ZiPS ERA Steamer ERA THE BAT ERA Mean StdDev
Génesis Cabrera 4.15 4.16 4.16 4.2 0.00
Graham Ashcraft 4.81 4.81 4.82 4.8 0.01
Aaron Nola 3.84 3.81 3.86 3.8 0.03
Emmet Sheehan 4.35 4.30 4.30 4.3 0.03
Joe Ryan 4.01 4.06 3.99 4.0 0.03
Alex Vesia 3.86 3.78 3.83 3.8 0.04
Scott Effross 4.00 4.07 4.09 4.1 0.05
Jackson Rutledge 5.25 5.35 5.28 5.3 0.05
Devin Williams 3.00 3.10 3.06 3.1 0.05
Gavin Williams 4.12 4.21 4.12 4.2 0.05
Spencer Strider 3.28 3.18 3.20 3.2 0.05
Andrés Muñoz 2.91 2.87 2.98 2.9 0.06
Zack Wheeler 3.61 3.54 3.49 3.5 0.06
Clarke Schmidt 4.41 4.38 4.50 4.4 0.06
Carmen Mlodzinski 4.27 4.39 4.35 4.3 0.06
Jason Adam 3.48 3.57 3.60 3.6 0.06
Chris Sale 3.84 3.74 3.86 3.8 0.06
Ryan Yarbrough 4.67 4.55 4.64 4.6 0.06
Taijuan Walker 4.67 4.75 4.80 4.7 0.06
Bryan Woo 4.11 4.02 3.99 4.0 0.07
Chase Silseth 4.36 4.34 4.47 4.4 0.07
Taj Bradley 4.18 4.04 4.09 4.1 0.07
Reese Olson 4.35 4.22 4.34 4.3 0.07
Joel Payamps 4.02 3.91 4.05 4.0 0.08
Lance Lynn 4.46 4.33 4.48 4.4 0.08
Carlos Estévez 4.25 4.30 4.41 4.3 0.08
Cole Ragans 4.04 3.93 4.08 4.0 0.08
Phil Bickford 4.22 4.35 4.20 4.3 0.08
Pablo López 3.68 3.52 3.63 3.6 0.08
Ross Stripling 4.45 4.49 4.61 4.5 0.09





Dan Szymborski is a senior writer for FanGraphs and the developer of the ZiPS projection system. He was a writer for ESPN.com from 2010-2018, a regular guest on a number of radio shows and podcasts, and a voting BBWAA member. He also maintains a terrible Twitter account at @DSzymborski.

37 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Jeremy Foxmember
2 months ago

Do projections for which all the projection systems agree have smaller errors than the mean of projections for which the various systems differ?

whatmonster
2 months ago
Reply to  Jeremy Fox

Seconding this question. Dan, I’d be very interested to see a historical test of how the projections for “agreement” and “disagreement” players performed against reality.

Doug Lampertmember
2 months ago
Reply to  whatmonster

Some of the reasons for the high variance cases are injuries and lack of professional data, which are going to make everyone worse. So I’d think that everyone does badly on most of the high variance players.

whatmonster
2 months ago
Reply to  Doug Lampert

Maybe. Probably! But let’s make Dan prove it!