5 Surprises in Minnesota’s 2021 ZiPS Projections

[Cody Christie]

As the calendar turns to a new year, it can be exciting to start thinking about the 2021 baseball season. Fans can hunt down different projection models to try and see how the Twins stack up against other top team’s in baseball. Last week, FanGraphs released their 2021 ZiPS projections for the Twins and there are a few surprises with some of the team’s top players.1. Healthy Arraez Heading for AL Batting Title

2021 ZiPS Projection: .313/.371/.406, 32 2B, 5 HR, 3.2 WAR

No. 1 Player Comp: Cecil Travis

Entering the 2020 season, Luis Arraez was coming off a tremendous rookie year and expectations were even higher for his sophomore campaign. His first 10 games were rough as he hit .212/.289/.502 without a single extra-base hit. He dealt with a knee injury throughout different parts of the season, but he seemed to put it all together over his final 22 games. During that stretch, he hit .367/.398/.481 with nine doubles and 12 runs scored.

 

Just like the 2020 projections, ZiPS pegs Arraez to lead the American League in batting average. His 3.2 WAR is also the highest on the team among position players as he finishes just ahead of Josh Donaldson, Nelson Cruz, and Max Kepler. It would certainly be exciting to have a healthy Arraez fighting for a batting title, but the Twins will likely want one of the other star players to lead the team in WAR.

 

2. Polanco Bounces Back

2021 ZiPS Projection: .279/.333/.440, 32 2B, 17 HR, 2.8 WAR

No. 1 Player Comp: Buddy Bell

There has been plenty of discussion this winter about what role Polanco should serve with the 2021 Twins. Will he be the team’s everyday shortstop, or does it make sense to bring in another option and shift Polanco to a utility role? During the last two off-seasons, Polanco has been forced to undergo ankle surgery and that’s a consideration for the team when planning for the future. Last year, Polanco posted career low marks in batting average, OBP, and slugging percentage.

 

Looking at ZiPS for Polanco and the projections clearly have him inline for a bounce back season. His projected slugging percentage would be six points higher than his career mark and his 17 home runs would only trail his 22 longballs in 2019. Also, he has accumulated 30 doubles or more in every season he’s played at least 130 games. Defensively, there were some improvements last year, but he has finished eighth among AL shortstops in SABR’s SDI in each of the last two seasons.

 

3. Maeda Set for Major Regression

2021 ZiPS Projection: 4.12 ERA, 135 1/3 IP, 154 K, 45 BB, 2.2 WAR

No. 1 Player Comp: John Montefusco

Maeda’s first season in a Twins uniform went about as well as it could possibly go. He finished runner-up in the AL Cy Young Voting after posting a 2.70 ERA, 161 ERA+ and a MLB leading 0.75 WHIP. It was everything the Twins hoped for when they traded for him and the best news is, he is under team control for the next three seasons. His season might have been the most dominant performance by a Twins starter since Johan Santana was traded away.

 

It seems highly unlikely for Maeda to be able to replicate his 2020 numbers during the 2021 campaign. The season will include more than 60 games and his 2020 totals were far superior to any previous season in his big-league career. ZiPS has his ERA 37 points higher than his career mark. Another oddity is that ZiPS has him scheduled to make eight appearances out of the bullpen, which would be similar to his time in Los Angeles. Maeda should outperform his ZiPS projections and Twins fans better hope he isn’t needed out of the bullpen.

 

4. Pineda Pitches Under 100 innings

2021 ZiPS Projection: 4.58 ERA, 92 1/3 IP, 84 K, 20 BB, 1.1 WAR

No. 1 Player Comp: Dave Eiland

Pineda’s time in Minnesota has been marked by one season where he was recovering from Tommy John surgery and parts of two seasons where he missed time due to a suspension. Last season, he made five starts and allowed 10 earned runs in 26 2/3 innings (3.38 ERA). Since joining the Twins he has posted a 1.16 WHIP and a 115 ERA+. The 2021 season can mark his first time pitching a full season for Minnesota, but the projections aren’t exactly kind to his performance.

 

Injuries have been part of Pineda’s professional career and that’s why ZiPS limits his projected innings pitched. In fact, there are over 10 pitchers projected to pitch more innings than Pineda for the 2021 Twins. His career ERA is 4.02 and he has only posted one season with an ERA higher than his projected 4.58. Another intriguing note is the fact Pineda can be a free agent following the 2021 season. Will he perform better in a contract year? Or will the Twins be willing to work out an extension?

 

5. ZiPS Loves Randy Dobnak

2021 ZiPS Projection: 4.53 ERA, 137 IP, 91 K, 37 BB, 1.6 WAR

No. 1 Player Comp: Dick Drago

Dobnak’s first two seasons in Minnesota have seen some ups and downs. Back in 2019, his rookie season was unbelievable as he posted a 1.59 ERA with a 1.13 WHIP across 28 1/3 innings. This culminated in the team turning to him for a Game 2 start in the ALDS. Last year, his ERA rose to 4.05 and he had a 1.35 WHIP while seeing his strikeout per walk total be cut in half. Eventually, he was optioned to the team’s alternate training site, but he was part of the team’s Wild Card roster.

 

In an absence of a minor league season, the minor league writers at Twins Daily held a minor league draft last summer. One of the biggest takeaways from that draft was how much ZiPS loves Randy Dobnak. His projected career WAR total was the highest in the draft and it helped Steve to walk away with the best overall team. Among pitchers, Dobnak is projected to have the team’s third highest WAR as he only trails Berrios and Maeda.

 

What other surprises were in the Twins 2021 ZiPS projections? Leave a COMMENT and start the discussion.

 

MORE FROM TWINS DAILY

— Latest Twins coverage from our writers

— Recent Twins discussion in our forums

— Follow Twins Daily via Twitter, Facebook or email

 

Click here to view the article

[sorney]

I kinda feel these are trash, because the 2020 season was so damn weird. Any data they used from 2020 to project anything has some wild variability. 

 

[DocBauer]

I agree it's hard to project anything for 2021 based on 2020 for all the reasons we already know.

 

I can completely understand upticks for both Polanco and Arraez as long as they are healthy. Clearly, they are betting on that.

 

I see no reason why Maeda couldn't slide a bit. Same with Pineda, but only to a small degree for both. Rocco does a great job of handling his starters, some would say he babies or under-utilizes them. And I get it. But I do appreciate his not insisting on overworking them and trying to get 200IP from everyone. With moderate rest, I see no reason Maeda and Pineda can't and won't be better than the listed projections.

 

Dobnak is interesting. Never expected him to be as good as his first 50 or so IP looked like, but after 75 IP we've seen a guy barely past rookie status who has largely performed well and shown some potential. If 2020 had been a full season, I have to wonder if he would have been an integral part of the 2nd half of the year after his demotion following a couple rough outings. So I can absolutely understand some optimism for a young back end starter who has flashed a bit.

[Trov]

How does ZIPS even work?  The Maeda projection seems the most odd.  Not saying he will perform to level of last year, but to project bull pen appearances is just odd.  I do not trust these at all as any kind of prediction.  It feels like they just look at player history, and generally career arcs of similar player histories based on age. 

[Major League Ready]

 

  On 1/5/2021 at 2:55 PM, Trov said:

How does ZIPS even work?  The Maeda projection seems the most odd.  Not saying he will perform to level of last year, but to project bull pen appearances is just odd.  I do not trust these at all as any kind of prediction.  It feels like they just look at player history, and generally career arcs of similar player histories based on age. 

 

https://blogs.fangraphs.com/the-2021-zips-projections-an-introduction/

[PDX Twin]

This seems like a simple extrapolation of recent records. Rather like the predictions of "big data" methods: decent as long as the future is just like the past. But it never is.

 

When Covid hit, many economic forecasters didn't know what to do because they had no past data from which to extrapolate. They actually had to think, a task at which some of them are inexperienced and, at times, incompetent.

[mikelink45]

It provides us with an off season talker.  Otherwise who cares?

 

[Doctor Gast]

W/ an upgrade at SS, I can see a very good year for Dobnak. A lot of grounders that got through last year won't in 2021 that is if Polanco is at 2B  

[jmlease1]

The biggest challenge with ZIPS is when you start drilling down into individual players. The Maeda example is perfect for this: regression towards the mean is pretty reasonable after last season's excellence...but projecting him to have ANY relief appearances shows they aren't accounting for his changed role with a new team. The Twins aren't going to push him into the bullpen absent some really strange circumstances, so this shows a flaw in their projection systems.

 

ZIPS also assumes that players with an injury history will continue to be injured when doing these projections; that's not an unreasonable assumption, but it's always a chancy thing.

 

I think the biggest surprises are a) how relatively unimpressed these projections are with Kirilloff and Jeffers; those ratings feel conservative to me, and again reveals one of the things I find the most frustrating about projection systems: they struggle to spot the young breakout guys. the projections on Astudillo, who I simply don't see getting anywhere near this much playing time in 2021, combined with the projection on Rortvedt. Clearly ZIPS likes Rortvedt's defense, but I'm not seeing him have this big of a role in 2021 (much as I like him) and Jeffers is going to get those ABs IMHO.

 

[Andrew Bryzgornia]

Some things I want to point out.

 

1) The player comps are who that player compares to at that same age in their career. So Polanco's Buddy Bell comp means he is most similar to Bell when Bell was 27 years old. 

 

2) 

  On 1/5/2021 at 2:55 PM, Trov said:

How does ZIPS even work?  The Maeda projection seems the most odd.  Not saying he will perform to level of last year, but to project bull pen appearances is just odd.  I do not trust these at all as any kind of prediction.  It feels like they just look at player history, and generally career arcs of similar player histories based on age. 

 

I don't recall where it was, but the ZiPS creator, Dan Szymborski, commented on FanGraphs relatively recently that projections are typically pessimistic towards pitchers simply because they get hurt so often. I personally prefer the Steamer projections, and they put many average pitchers at the 4.50 - 5.00 ERA range, mainly because the league average ERA is around 4.50. (For example, they have Maeda at a 4.43 ERA in 2021). Even with the 4.12 ERA projection from ZiPS, that's still better than the 2020 league average ERA of 4.45. I'm assuming it's because the majority of his career was in the NL, which has pitchers hitting, so there should be regression simply from the change in leagues and having to face DHs. But, also, because pitchers get hurt. 

 

Additionally, yeah, the bullpen appearance projection for Maeda is a little goofy since he's no longer in LA, but that's one of those things that a computer isn't going to know. The algorithm sees Maeda as a sorta-swingman when really he's now a full-time starter. Szymborski should probably make an edit for Maeda, but then we get into the territory of arguing that edits should be done for other players too, and then you're moving away from the point of a computer projection, which is supposed to be fully objective.

 

3) Don't take projections that seriously. It's like meteorology, it's just an educated guess using past experience to attempt to predict the future. It's best to use multiple projection systems to get an overall idea, but it's even better to not get worked up about anything that looks wrong. 

[Andrew Bryzgornia]

 

  On 1/5/2021 at 4:58 PM, JLease said:

The biggest challenge with ZIPS is when you start drilling down into individual players. The Maeda example is perfect for this: regression towards the mean is pretty reasonable after last season's excellence...but projecting him to have ANY relief appearances shows they aren't accounting for his changed role with a new team. The Twins aren't going to push him into the bullpen absent some really strange circumstances, so this shows a flaw in their projection systems.

 

ZIPS also assumes that players with an injury history will continue to be injured when doing these projections; that's not an unreasonable assumption, but it's always a chancy thing.

 

I think the biggest surprises are a) how relatively unimpressed these projections are with Kirilloff and Jeffers; those ratings feel conservative to me, and again reveals one of the things I find the most frustrating about projection systems: they struggle to spot the young breakout guys. the projections on Astudillo, who I simply don't see getting anywhere near this much playing time in 2021, combined with the projection on Rortvedt. Clearly ZIPS likes Rortvedt's defense, but I'm not seeing him have this big of a role in 2021 (much as I like him) and Jeffers is going to get those ABs IMHO.

 

1) If a projection specifically focused on just the prior year, it would be a bad projection. It has to use multiple years of data, which is why we get that noise with the projected bullpen appearances.

 

2) Szymborski says at the end of every ZiPS post on FanGraphs, "ZiPS is agnostic about future playing time by design." 

[Shaitan]

Dobnak is the only surprise to me.

 

And that they think .313 wins a batting title.

[ashbury]

  On 1/5/2021 at 7:05 PM, Shaitan said:

And that they think .313 wins a batting title.

That's probably the indication you're misunderstanding the meaning of their projections.

 

Each number (BA, etc) represents something of a middle spot within a whole cloud or range of possibilities for that one player. Better hitters have a higher middle spot. But it's a certainty that some players will overperform ("career year") and others will underperform ("slump", or obvious injuries).

 

Someone will hit .330 or above, almost certainly.

 

If .313 is the highest middle spot for batting average, it likely means they think that Arraez is the best bet to reach that .330+ level.

[Channing1964]

It's fun to talk about now but it's all b.s. when the games really start

[jmlease1]

 

  On 1/5/2021 at 6:57 PM, Andrew Bryzgornia said:

1) If a projection specifically focused on just the prior year, it would be a bad projection. It has to use multiple years of data, which is why we get that noise with the projected bullpen appearances.

 

2) Szymborski says at the end of every ZiPS post on FanGraphs, "ZiPS is agnostic about future playing time by design." 

 

1) sure, but it should also account for changes in role. but ZiPS doesn't appear to have that ability.

2) Is it, though? It does account for injuries by players: guys who have been historically healthy and playing full time get projected to do the same, and guys who haven't been healthy don't get projected to play a full season. So how does a projection for someone like Rortvedt work? He's never gotten an AB in MLB, has 2-4 players ahead of him on the roster (depending on what you think of Astudillo as a catcher and the AAA guy they signed)...why would he get projected to get the same level of ABs as Jeffers? And the idea that Rortvedt is a significantly better defender at this point and that his offense would be fairly close to Jeffers so that he'd be more valuable is a bit silly. I'm a huge Rortvedt guy and even I know that  Jeffers is far past him right now.

[Vanimal46]

A nice surprise is seeing 12 batters projected with a >100 OPS+. A lineup full of league average or better hitters is going to be fun to see!

[old nurse]

Projections as a team are far more accurate than for individuals