Notes from the 2021 ZiPS projections

Over at Fangraphs earlier this week Dan Szymborski posted the 2021 Twins projections from his proprietary ZiPS projection system.

If you aren’t familiar with ZiPS or are interested in a more detailed explanation of how the system works, click here.

While ZiPS is just one of many projection systems out there, it is one of the most prominent and Szymborski adds new features based on research every year to improve the system’s accuracy. That means the system generally keeps up with the new data and analytics made available, although he points out that he takes a conservative approach to his model design and data is only included once there is confidence it improves accuracy.

There is no such thing as a “right” projection or a “wrong” projection and the bottom line projections published or publicly available for download most often represent mid-points (meaning most likely) of a range of possible outcomes. For example, the projections available on Fangraphs are medians – implying approximately 50% of the actual outcomes will be worse than projected and approximately 50% of the actual outcomes will be better than projected.

I believe wholeheartedly in the old saying in statistics that “All models are wrong, some are useful,” but also find it fun to peruse the projections for interesting nuggets or to unapologetically cherry pick data that confirms or rejects something we believe to be true or false of our favorite team and players.

With that in mind, here are some interesting things I saw in the initial projections for the Twins.

Keep adding to the roster

The Twins roster, as it was constructed on Monday, is projected to be slight underdogs to Chicago in the AL Central. The offseason roster building season is far from over and it seems likely the Twins will add more before the season starts, but it seems clear that taking the AL Central for a third straight season is anything but guaranteed.

It seems as though Chicago is the main (only?) competition. Cleveland appears willing to take a step back (and may trade star shortstop Francisco Lindor) to save money, while Kansas City and Detroit continue to progress through various stages of rebuilding. Despite that, the Royals seem to be making some shrewd offseason moves and may be slightly better than we expect in the end and sleeping on Cleveland seems unwise given their proven ability to develop dominant pitching.

Nonetheless, the case seems clear that Minnesota needs to keep strengthening the roster for 2021. Szymborski also ran the projections for the scenario where Nelson Cruz returns as the Twins’ DH in 2021 and that resulted in the Twins coming out slightly on top of the AL Central standings.

The offense projects to be in slightly better shape than the pitching staff and is expected to rank somewhere around 4th to 8th in runs scored. Szymborski described the pitching staff as average and in need of more depth. The addition of Hansel Robles is one move to that end but the team likely needs more — both in the rotation and the bullpen.

Luis Arraez, batting champ (again)

While 2020 was injury marred, Luis Arraez continued to demonstrate he has an elite hit tool and ability to hit for average. Over 121 plate appearances Arraez hit .321 / .364 / .402, not quite the same production as his 2019 rookie campaign, but still strong. Arraez still hasn’t accumulated a full season’s worth of plate appearances in his career (487) but is carrying a .331 batting average and 8.2% strikeout rate when the respective league averages are less than .250 and just above 23%.

ZiPS projected Arraez to lead the majors in batting average last season and is doing so again for 2021. With a .313 average and more walks than strikeouts, Arraez takes home the highest position player WAR projection (3.2).

Projecting Nelson Cruz

While we wait for Nelson Cruz and the Twins to find an agreement to bring him back for another season, the projections offer a glimpse as to why the decision to bring him back isn’t so simple.

ZiPS is bullish on Cruz remaining productive in a season during which he’ll turn 41. It’s projecting a .276 / .361 / .550 triple slash line and 3.1 WAR, numbers that would reflect a slight decline from the heights of the past two seasons but that still remain the best composite offensive production on the team by a significant margin.

Steamer projections, on the other hand, have a more pessimistic view and project .256 / .344 / .490 and just 1.1 WAR. The ZiPS line above works out to about 40% better than average, whereas the Steamer projection is more like 15% above.

For a soon to be 41 year old those numbers are fantastic, and both lines would be useful additions to the Twins’ 2021 roster. But the disparity between the two systems is an example of uncertainty that comes with Nellie’s profile — there just isn’t much historical comparison to use to get a good sense of what to expect.

Kiriloff vs. Rosario

One of the nice things about the ZiPS projections is that players are projected in the batches with their most recent team, even if they aren’t part of that club anymore. For us, that means its easy to make a comparison of Alex Kiriloff and Eddie Rosario.

Kiriloff — .287 BA / .330 OBP / .460 SLG, 111 OPS+, -2 DEF, 1.5 WAR

Rosario — .277 BA / .314 OBP / .488 SLG, 112 OPS+, 0 DEF, 2.0 WAR

ZiPS gives Rosario slight edges in slugging and on defense, while giving Kiriloff the nod in batting average and on-base percentage. However you want to slice it, the projections between these two players are pretty similar for 2021. ZiPS is intentionally not a playing time projector, so I’m going to avoid including any counting stats that may be skewed by varying playing time estimates. Certainly the projections above are close enough to be within the models’ margins of error and I think they help to illustrate why the Twins felt they could release Rosario. As has been covered ad nauseum already this winter, Kiriloff projects to provide nearly the same production but for a fraction of the cost.

Should Kiriloff not live up to expectations the Twins are in great shape for outfield depth. Jake Cave (103 OPS+) and Brent Rooker (98 OPS+) also project to be about league average producers offensively.

Rotation has holes

ZiPS projects Kenta Maeda and José Berríos as far and away the Twins’ top starting pitchers with 3.91 FIP / 93 ERA- and 3.95 FIP / 88 ERA-, respectively. After them, ZiPS sees the Twins’ starting options as a whole bunch of average or somewhat worse options which further highlights the need for reinforcements to be added this winter. Those options include:

Randy Dobnak — 4.47 FIP / 102 ERA-

Michael Pineda — 4.36 FIP / 103 ERA-

Lewis Thorpe — 4.73 FIP / 109 ERA-

Devin Smeltzer — 4.84 FIP / 113 ERA-

From a depth perspective, the Twins are in decent shape as both Bailey Ober (108 ERA-) and Jhoan Duran (112-) project comparably with the group broken out above. It’s not so much that the Twins are suffering from a lack of passable options for the starting rotation as it is that the options may not have the upside and impact a contender would need.

Players ZiPS doesn’t believe in

One of the projections I was most interested to see was for Ryan Jeffers. Jeffers more than held his own with the bat and in the field as a rookie last season, despite having only played 29 games above A-ball in his minor league career. In only 62 plate appearances, Jeffers hit .273 / .355 / .436 and was about 19 percent above league average by wRC+. That ranked 13th out of the 58 catchers that had at least 50 PAs last season. ZiPS isn’t buying that level of offensive performance, however, projecting .239 / .305 / .385 and 85 OPS+ in 2021.

On the mound, ZiPS can also be counted among the doubters of reliever Jorge Alcala. Despite his power arsenal and rookie season success, ZiPS sees Alcala with 5.20 ERA / 5.12 FIP and a 10.9% walk rate next season. Some of this could be due to the projections including Alcala spending some time as a starting pitcher and working more than 90 innings in total. It would be interesting to see if his projection would have improved if the models assumed he would work exclusively in a relief role.

ZiPS is also pessimistic about Cody Stashak, assigning him 4.83 ERA / 4.70 FIP and 109 ERA- due in some measure to a projected rate of 1.7 home runs allowed per nine innings. That’s one of the worst projected rates on the team and also quite different than Stashak’s career rate thus far (1.13 HR/9 in 40 career innings).

Prospect Nick Gordon is another player that ZiPS doesn’t give a rosy outlook. An option for the utility player roles currently open due to Marwin Gonzalez and Ehire Adrianza heading to free agency, Gordon is projected to produce more similarly to Adrianza (82 OPS+) than Marwin (97 OPS+). ZiPS projects .252 / .296 / .370 and just 79 OPS+ for Gordon in 2021. On the brighter side, thanks to his expected solid defense and baserunning, Gordon’s WAR projection is equal to that of Gonzalez (1.1 WAR).

Fun with Comps

In addition to projecting statistics for the upcoming season, ZiPS also compares player’s performance and to the baseline of every player in its database (which includes every Major Leaguer since the deadball era) at every point in their career. This helps to inform future projections through probable aging curves while also providing each player with their most similar comparable players from baseball history. The comps reflect the best match for each player at a similar point in their careers.

Here are a few of the comps that I thought were interesting:

José Berríos as Chien-Ming Wang — Through about four full MLB seasons and just more than 600 innings pitched, Berríos has posted 4.19 ERA / 4.09 FIP and 11.1 fWAR while the Yankee Wang’s numbers at a similar point were 3.79 ERA / 3.90 FIP and 10.4 fWAR. In that light, Berríos’ match with Wang makes sense. After that stretch to begin his career, Wang struggled to stay healthy and performed below replacement level for the remainder of his career. Let’s hope for Berríos’ sake that this comp doesn’t make sense for long.

Trevor Larnach as Andre Ethier — Both are left-handed hitting outfielders drafted from PAC-12 baseball powerhouses and I think nearly everyone would be thrilled if Larnach’s (Twins #3 prospect) career turned out similarly to Ethier’s. Ethier played parts of 12 seasons with the Dodgers and finished his career as a .285 hitter who got on base nearly 36% of the time and bopped 162 homers on the way to producing 24.2 fWAR.

Gilberto Celestino as Magglio Ordonez — Celestino (Twins #14 prospect) was part of the Twins’ return (along with Alcala) for trading Ryan Pressly to Houston in 2018. Ordonez, who was a very poorly regarded defensive outfielder in his otherwise successful career, makes for a strange comp given the scout evaluations of Celestino as a plus defender in the outfield.

Ben Rortvedt as Henry Blanco — This one might be too perfect. Added to the 40-man roster earlier this winter, Rortvedt (Twins #26 prospect) profiles best as a glove-first backup catcher. He has excellent blocking and receiving skills and gets high marks for his ability to frame pitches and work with a staff, according to the MLB pipeline scouting report. That sounds an awful lot like Henry Blanco, doesn’t it?

More scrap heap treasures?

One of the things the Twins have developed a penchant for in recent seasons is turning low-cost pitching acquisitions into valuable contributors. They will try to do so again in 2021, having already added right-handers Ian Gibaut (waivers) and Derek Law (minor league deal).

ZiPS sees both of these low-risk fliers as potentially useful contributors. The 30 year old Law is projected for 4.36 ERA / 4.32 FIP, more than a strikeout per inning, and 99 ERA-. Gibaut, 27, is projected for 4.50 ERA / 4.61 FIP, more than a strikeout per inning, and 102 ERA-.

Near league average relief production out of a minor league free agent and a waiver claim would certainly be considered wins in Minnesota.

---

John is a contributor to Twinkie Town with an emphasis on analytics. He is a lifelong Twins fan and former college pitcher. You can follow him on Twitter @JohnFoley_21.