Looking Back at the 2022 ZiPS Projections

© Charles LeClaire-USA TODAY Sports

Before we get to the 2023 ZiPS projections, there’s nonetheless some unfinished work from 2022 to do. Looking at which projections went the most wayward is certainly an train that makes my soul cringe a bit, however in any mannequin, being fallacious is a vital element of ultimately being proper. Calibration is a long-term challenge, and whereas chasing higher accuracy in imply projections is not prone to lead to any big bounty — there is a cause projection methods are so tightly clustered — there’s nonetheless enchancment available in issues like calibrating uncertainty and lengthy -term date

Let’s begin with how groups carried out versus their projections:

2022 ZiPS Projected Wins vs. Actual Wins

Team CiPS Real Difference
Baltimore Orioles 64 83 19
Los Angeles Dodgers 93 111 18
Houston Astros 90 106 16
Cleveland Guardians 78 92 14
New York Mets 88 101 13
New York Yankees 88 99 11
Atlanta Braves 90 101 11
Seattle Mariners 85 90 5
St. Louis Cardinals 88 93 5
Philadelphia Phillies 83 87 4
Toronto Blue Jays 88 92 4
Arizona Diamondbacks 71 74 3
San Diego Padres 89 89 0
Milwaukee Brewers 87 86 -1
Tampa Bay Rays 88 86 -2
Colorado Rockies 70 68 -2
Chicago Cubs 77 74 -3
San Francisco Giants 85 81 -4
Kansas City Royals 70 65 -5
Pittsburgh Pirates 68 62 -6
Chicago White Sox 88 81 -7
Detroit Tigers 73 66 -7
Minnesota Twins 86 78 -8
Los Angeles Angels 81 73 -8
Oakland Athletics 68 60 -8
Texas Rangers 77 68 -9
Boston Red Sox 88 78 -10
Cincinnati Reds 74 62 -12
Miami Marlins 82 69 -13
Washington Nationals 76 55 -21

Teams have gotten a bit extra polarized in how they’re run in-season. Looking at the in-season ZiPS projections, roster power has diverse rather more in recent times than once I began doing this. It would not be stunning to see the imply absolute error — for an train like this, I wish to use the easiest device that will get the level throughout — creep up over time. That is the case right here, as the MAE of 8.3 wins is above the ZiPS historic common of seven.5 (not together with 2020). ZiPS underperformed its traditional matchup vs. Vegas, solely going 17-13 in over/unders as of the date of launch (April 6); traditionally, ZiPS has averaged 19-11.

What would an ideal projection appear to be? If you knew the win likelihood of each sport performed earlier than it was performed, the MAE of your seasonal projections, based mostly on the hypergeometric distribution, can be round 5 wins.

Thirty-six % of the estimated error in ZiPS was merely guessing fallacious on roster building. With excellent data of who truly performed, the MAE drops to 7.1 wins. The greatest good points in accuracy would have been the Boston Red Sox (88 wins projected, 78 realizing plate appearances/complete batters confronted), the New York Mets (88, 96), and the Cleveland Guardians (78, 84). Amusingly, the Orioles had been one among the 4 groups ZiPS would have truly missed on by a bigger quantity if the system had recognized taking part in time data forward of time. Simply put, ZiPS would have been even purple unfavourable about the O’s if it had recognized what their rotation would appear to be!

With 10 deciles and 30 groups, you ideally need three groups to complete with a percentile projection in every decile. In different phrases, you need three groups to fail to fulfill their Tenth-percentile projection, three groups to at least hit their Ninetieth-percentile projections, and so forth. So how did ZiPS do?

2022 CiPS Projection Deciles

Decile Teams
First 3
Second 3
Third 5
Fourth 3
Fifth 3
Sixth 1
Seventh 3
Eighth 2
Ninth 2
Tenth 5

Amusingly, ZiPS acquired 4 of the six divisions in the proper order, with the AL Central error being Cleveland truly ending first as a substitute of third, and an actual mess in the AL East (4 groups projected at 88 wins and the O’s had been the greatest out – performers in baseball). There was no change in the year-to-year correlation between errors; there is not any vital relationship (an r-squared of -0.0013) over the entirety of ZiPS’ historical past. This can also be true on a per-franchise stage. What this implies is that there is no such thing as a tendency for errors one yr to foretell errors the following yr. This holds true even on the workforce stage, with the Mariners having the highest r-squared in season-to-season errors at 0.02. So, no, there is not any cause to suppose the 2021 Giants found a particular potential to outperform projections; they merely outperformed projections in 2021.

For hitters and pitchers with at least 300 plate appearances or batters confronted, the MAE for OPS+ was 16.3 factors, whereas for ERA+, it was 20.9 factors. Both of those are in keeping with the outcomes over the final decade (16.4 and 21.1, respectively). I truly anticipated to perform a little worse than regular.

Here are the hitters the system most underestimated:

Biggest ZiPS Hitter Misses (Positive)

I imagine that the miss for Albert Pujols is the greatest ever in ZiPS for a participant 35 years or older. And I’m very comfortable to have missed on that one! I believe it is unlucky that we’ve a participant who’s a slam-dunk, first-ballot Hall of Famer, however who mainly spent the final decade of his profession hitting like Darin Ruf as a substitute of Jimmie Foxx. There are loads of youthful baseball followers whose solely expertise of Pujols has been that of a mediocre 1B/DH. While 2022 would not problem any of the basic seasons from Pujols’ first decade of play, getting one final burst of glory was actually cool to see and lifted his farewell season above the unhappy trudge of Derek Jeter’s limitless goodbye in his last marketing campaign.

The unexpectedly wonderful efficiency of Michael Harris II had a higher influence on his future projections than any of the different prime performers right here. Given his younger age, there was loads of uncertainty about his upside. He was truly significantly extra prone to hit his precise OPS+ (135) than any of the different 14 gamers listed. Now that he is apparently hitting that upside, his 2023 ZiPS projection has seen an enormous bump. With solely 154 plate appearances, Matt Carpenter didn’t make this record, however he was the most surprising participant total in 2022, with the projections solely giving him a one-in-2,300 likelihood of hitting in addition to he did, even in a comparatively small pattern.

Now listed below are hitters who most underperformed their projections:

Biggest ZiPS Hitter Misses (Negative)

Three White Sox made this record, with two of these three anticipated to be amongst the workforce’s most precious gamers heading into the season (I’ll allow you to guess the odd man out). It’s not stunning Chicago’s offense struggles to attain runs in 2022. One attention-grabbing factor from the bigger dataset, which you’ll see in a few of the names above, is that being a Three True Outcomes participant had a slight impact on projection accuracy in 2022, one which hadn’t been current at any time in the previous. I’m positively curious to see if this improvement persists, assuming we’ve the similar deadened baseballs we appeared to have this season. Now let’s flip to the pitchers:

Biggest ZiPS Pitcher Misses (Positive)

I believe half of the Dodgers beginning rotation made this record, which is hardly stunning provided that the Dodgers outperformed their already wonderful projection by just one fewer win than the Orioles did. Andrew Heaney wasn’t far off from making this group, both, and that is with ZiPS already being extra optimistic about the left-hander than most. If you wager on Matt Moore being a extremely priceless aid pitcher, pat your self on the again; ZiPS certain did not! Now for the under-performers:

Biggest ZiPS Pitcher Misses (Negative)

Hey look, extra White Sox! Given how poorly a lot of their workforce did relative to the projections, I’m truly a bit shocked that ZiPS solely over-projected them by seven wins (88 vs. 81). I agreed with ZiPS that Dallas Keuchel would at least be a serviceable fifth starter this season, which completely didn’t occur! That mentioned, I did not agree with ZiPS’ comparatively pleasant projection for Patrick Corbin, so take it thatset of algorithms I developed!

Let’s end up with the calibration information for pitchers and hitters. This is definitely one thing I concentrate on much more than the imply projections. Models aren’t proper, they’re helpful, and a correctly calibrated projection system ought to know the way usually it’s anticipated to be fallacious.

2022 CiPS Projection Deciles

Decile ERA+ OPS+
First 10.7% 8.3%
Second 10.1% 10.5%
Third 8.4% 9.4%
Fourth 11.8% 10.5%
Fifth 11.2% 9.8%
Sixth 10.7% 11.2%
Seventh 9.6% 8.7%
Eighth 8.4% 10.5%
Ninth 7.9% 12.0%
Tenth 11.2% 9.1%

Next up: the comically preliminary ZiPS projected workforce standings for 2023!

Leave a Comment