The ZiPS Projections Midpoint Roundup of Triumph and Shame: The Players

As we’ve passed both the literal midpoint of the season (1,215 games) and the philosophical one (the All-Star Game), it’s time to look back at the sample-size-fueled joy and sadness of the best and worst individual player projections for the first half. Projecting this season was particularly tricky for the obvious reason that the 2020 season was only 60 games and there was no minor league season at all. There’s not really much you can do to compensate for the lack of data; in the end, you’re likely to be less accurate no matter your approach. This isn’t as big a deal when it comes to the team projections, where there are enough players that you just hope the mistakes aren’t all in the same direction, but it can matter when you’re talking about a single player. (In hindsight, it makes me happy I never needed to project the 1995 season using replacement players, if that had actually happened.)

Let’s start with the hitters. Given the volatility of defensive measures, this is a comparison of the offensive numbers, not WAR as a whole. I’m setting 200 plate appearances as the minimum. Here are the position players ZiPS most underrated:

ZiPS Projections – The Most Underrated Hitters
Name wRC+ Preseason wRC+ Difference
Buster Posey 164 89 75
Vladimir Guerrero Jr. 189 120 69
Shohei Ohtani 180 116 64
Brandon Crawford 147 84 63
Cedric Mullins 151 90 61
Akil Baddoo 121 61 60
Bryan Reynolds 146 96 50
Mike Zunino 123 75 48
Tyler O’Neill 138 90 48
Tyler Stephenson 122 74 48
Adolis García 127 81 46
Nick Castellanos 156 111 45
Omar Narváez 137 94 43
Yuli Gurriel 136 93 43
Jonathan India 123 81 42

A few of the names on the list don’t really surprise me. It was difficult to get a read on what Shohei Ohtani was going to do coming into the season, both because his professional career in the United States has been so short and because recovering from a pitching-related injury introduced an additional complication. I’m actually quite happy that ZiPS missed on this one; Ohtani is having the kind of year I’ve wanted to see since the moment it was clear that he was going to come over to the US.

Meanwhile, there being a number of young players among the biggest misses isn’t unusual. Even if you somehow knew that a player who is early in his career was going to break out, it wouldn’t mean you’d necessarily know quite when. Mean projections are supposed to get a lot wrong, so seeing Vladimir Guerrero Jr. and Akil Baddoo and Jonathan India here is not particularly odd. Guerrero was the top pick on my breakout hitters list, so I’m not all that upset about the ZiPS mean projection being on the light side.

Then there’s Mike Zunino, though honestly, I think I’d be more surprised if ZiPS had gotten Zunino right than wrong. He has an 86 career wRC+, but over his nine seasons in the majors, his average deviation from his career average is a whopping 25 points of wRC+. The “typical” Zunino season barely exists.

That two of the biggest misses were Buster Posey and Brandon Crawford goes a long way to explaining how the Giants are second in the National League in runs scored. Usually, this type of player — the early-to-mid 30s veteran gently declining — is one of the easier archetypes to project, but in this case, they’re suddenly matching their best performances ever. And in Crawford’s case, he’s greatly exceeding them!

Next, we’ll consider the most overrated hitters:

ZiPS Projections – The Most Overrated Hitters
Player wRC+ Preseason wRC+ Difference
Jorge Soler 66 124 -58
Marcell Ozuna 77 132 -55
Gleyber Torres 82 134 -52
Hunter Dozier 58 110 -52
Eugenio Suárez 68 114 -46
Jackie Bradley Jr. 49 93 -44
Anthony Rendon 98 142 -44
Ha-Seong Kim 김하성 71 110 -39
Kevin Newman 45 84 -39
Marwin Gonzalez 61 98 -37
Michael Conforto 94 129 -35
Anthony Santander 77 112 -35
Ian Happ 78 113 -35
Christian Walker 69 102 -33
Miguel Sanó 92 125 -33

There are a lot more surprises on this list given that we’re primarily talking about well-established hitters in the middle of their careers. Jackie Bradley Jr. is the only hitter here who I had pegged to underperform his projection coming into the season, though the possibility existed with Kim since we only had KBO stats to go on.

I’ll beg out and blame injuries for Anthony Rendon, but these players have otherwise been relatively healthy. Gleyber Torres is particularly confounding. At age 21 and 22, he showed the ability to turn on a pitch with authority, posting ISO numbers above .200 in both seasons. But that ability seemed to disappear in 2020 and especially ’21. It’s not just that making more contact has resulted in more weak hits off borderline pitches; he’s also struggling with pitches in the heart of the strike zone. There have been 117 players who have seen at least 300 pitches that Statcast defines as in the “heart” of the strike zone this year. Torres ranks 105th in slugging percentage and 96th in exit velocity, with his .431 SLG on those pitches paling in comparison to his .691 in 2018 and .776 in ’19. This kind of decline is disappointing but not completely unheard of: Randy Ready and Chris Speier are high on Torres’ ZiPS comp list, among much more distinguished players. ZiPS has Torres rebounding to a 120 wRC+ in 2022, but at this point, I have to take the under.

Overall, the average miss on the hitters has been 18 points of wRC+. Normally with the percentage of the season played so far, that mark is around 16 or 17 points, so the projections haven’t missed by as much as I expected.

Now to the pitchers, starting with the most underrated:

ZiPS Projections – The Most Underrated Pitchers
Name ERA- Preseason ERA- Difference
John Means 53 110 57
Kyle Gibson 53 108 55
James Kaprielian 70 122 52
Danny Duffy 57 109 52
Carlos Rodón 54 105 51
Cristian Javier 72 122 50
Kevin Gausman 44 93 49
Lance Lynn 46 89 43
Trevor Rogers 60 99 39
Jacob deGrom 28 67 39
Zack Wheeler 56 92 36
Wade Miley 68 104 36
Taijuan Walker 65 101 36
Spencer Turnbull 65 100 35
Anthony DeSclafani 69 104 35

As I’m from Baltimore and also had him on my pitcher breakouts list, I’m not at all displeased to see John Means make this list. Shoulder injuries are always concerning, but I’m not as worried as some that his 4.19 FIP portends regression. He’s been lucky in his BABIP, but the argument can be made that he’s been unlucky in a couple of the FIP components. Based on the peripheral numbers, ZiPS sees Means as the second-biggest strikeout underperformer compared to his zSO, behind only Matthew Boyd. In swinging strike rate, Means ranks 17th in the majors, sandwiched between Ohtani and Gerrit Cole, but only 43rd in strikeout percentage.

After him spending a half-season fighting Bob Gibson’s modern ERA record, I’d have been extremely concerned about calibration issues if the mean projection for Jacob deGrom was actually correct. Something like this should never be the over/under.

Carlos Rodón is by far the biggest surprise for me here. Yes, he’s been battling injuries for years, and it’s always tricky to evaluate a pitcher in that situation. But if you told me he’d get back just to 2015-2016 levels of performance, I’d still have been surprised. Of course, Rodón didn’t even stop there; he’s throwing 96 mph and looking a lot like the best-case scenario the team envisioned when the White Sox took him with the third pick in the 2014 draft. And he’s not even really beating his seasonal FIP! Coming into 2021, ZiPS projected Rodón for a 4.68 ERA for 2022. That’s down to a 3.31 now, the largest improvement in baseball.

Now for the most overrated hurlers:

ZiPS Projections – The Most Overrated Pitchers
Name ERA- Preseason ERA- Difference
Matt Shoemaker 184 111 -73
Dylan Bundy 155 96 -59
Carlos Martínez 160 104 -56
Jake Arrieta 157 104 -53
Justus Sheffield 155 105 -50
David Peterson 145 98 -47
José Ureña 146 100 -46
Riley Smith 152 106 -46
Chris Paddack 137 92 -45
Marco Gonzales 141 96 -45
Blake Snell 127 83 -44
Patrick Corbin 131 89 -42
Matt Harvey 178 137 -41
Brad Keller 135 99 -36
Andrew Heaney 125 89 -36

Considering that ZiPS projected Matt Harvey for a 6.08 ERA, it’s a bummer to see him on this list.

I remain hopeful about Chris Paddack, though the membership roll for the Paddack Optimists Club appears to be shrinking. Some of the players on this list are playing poorly, but in Paddack’s case, his FIP is basically identical to what it was in his rookie season. Another bounce-back candidate is Dylan Bundy. When I posted the pitcher zStat leaderboards back in May, Bundy was one of the biggest FIP underperformers. Some of the other underperformers have already salvaged their seasons; Luis Castillo and Germán Márquez have been terrific since then. But neither Bundy nor teammate Andrew Heaney have turned it around.

On the other hand, Jake Arrieta is one pitcher I don’t expect to recover. Calling his command spotty would be overly generous at this point, and as they did in 2020, hitters are successfully waiting him out until they get a crushable offering. Nor am I optimistic about Patrick Corbin. He lost 100 rpm on his slider in 2020 before the crackdown on sticky substances, and the probability a hitter would whiff on one dropped accordingly. His average spin rate has dropped from 2241 rpm in 2020 to 2210 this year, and in the running average of his 50 most recent sliders, it has dropped to under 2200 in the last month (currently 2171).

Looking at the projections as a whole, I’ve examined how errors are correlated, hoping it would give me an idea of whether there were specific types of players who the odd 2020 season hurt more than others or at least made less predictable. So far, I’ve found nothing systematic along those lines. A wider post-mortem will be performed after the season, and hopefully, it will offer some additional insight about how to project two of the weirdest years in baseball history.





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.

newest oldest most voted
sadtrombone
Member
Member
sadtrombone

All of these guys qualify as legitimate surprises except Gleyber Torres. He suffers a bit from “Bryce Harper Syndrome,” which is what happens when a player has a fantastic season at a particularly young age and the projection systems can’t make sense of him anymore. Harper is the most extreme example of this and no one else comes close, because he has 4-win seasons at age 19 and 20, a bonkers 9 win season at age 22, and then went right back to 3-5 win seasons. Projection systems have no idea what to do with that when a player is this good at young ages, and continue to project them to continue building on their success after they’ve stalled out.

Gleyber Torres isn’t anywhere near that level of craziness because he has never been that good, but projection systems saw him whack 38 homers at age 22 and the computers can’t fathom why he can’t at least get back to that level. And it is really weird, because he’s on pace for like a six-homer season. You can’t exactly blame ZiPs for that. But one thing to note here is that his xwOBA has actually been remarkably consistent across all of those years. He’s a little under that now but by xwOBA he’s probably more like a league average hitter right now, like what he did last year, and what he did his first two years was also more like a league average hitter as well. IIRC ZiPs does take batted ball data into account but the results were so good so young I guess it threw it for a loop anyway.

Cheeknbut
Member
Cheeknbut

Sorry, but just looking at his xwOBA over that timeframe is not a great take. Sure, his xwOBA is only down around .010-.015 points from his 18-19 years, but his xwOBACON is down .050-.075 points since then. It’s just that he’s also cut his strikeouts and upped his walks significantly starting in 2020.

Travis L
Member
Member
Travis L

Projection systems can handle outliers just fine using regressions. How did you reach this conclusion about a Harper season? It doesn’t match with my experience in building models.

sadtrombone
Member
Member
sadtrombone

Go back and take a look at Harper’s projections over the last few years. Took something like 5 years to get back to the right number. To be fair, it’s tough to handle a guy who is that successful that young–it’s just such an absurd number. But I think there’s an increasing number of cases (Harper, Lindor, Machado, to a lesser extent Juan Soto) of guys who come up and are playing like 10 year veterans and we expect them to somehow improve. And sometimes they do, and sometimes they don’t.

jdr
Member
jdr

David Wright is another dude who came up fully formed and then kinda bounced around depending on health.

tung_twista
Member
tung_twista

Let’s do take a look at how Harper performed compared to the projection (OPS)

2015 zips .859 Harper 1.109
2016 zips .928 Harper .819
2017 zips .927 Harper 1.008
2018 zips .967 Harper .889
2019 zips .944 Harper .882
2020 zips .909 Harper .962
2021 zips .914 Harper .891

We all know the ridiculous year Harper had in 2015 and the subsequent extremely disappointing 2016.
If you look at his record since then, it just looks like Harper is alternating on each sides of expectation as most players do.
I don’t see much to support your ‘Harper broke the projection system’ narrative.
None of the projections seem ridiculous given the available information at the time.

mchilcott
Member
mchilcott

According to those numbers above Harper has under-performed his projected OPS by a mean of .023 since the breakout 2015. Seems well within the standard error bars to me.

dukewinslow
Member
Member
dukewinslow

Give the sd right now is basically 0.2, he’d have to play a pretty long time for that to be a significant underperformance (like, a century, I can’t do the power calc in my head or on the phone and the error is correlated so it’d be wrong anyway)