Sports News

zStats for Pitchers, June Update

Kirby Lee-USA TODAY Sports

Among the statistics created by Statcast and similar tracking tools in recent years is an entire category of statistics sometimes called “expected statistics.” These types of numbers evoke mixed feelings among fans – especially when they suggest their team’s best player is overachieving – but they serve the important purpose of linking Statcast data to events on the field. Events in baseball, whether it’s a single or a homer or a strikeout or whatever, happen for reasons, and this kind of data allows us to better evaluate baseball at the first level.

Although the lucky home run or single you see still counts in the scoreboard and box score, the expected statistics help us in the next reveal. Naturally, as the developer of the ZiPS prediction tool for 20 years (!), I have a great interest in improving these predictions. Statcast has its own way of measuring expected stats, which you’ll see everywhere with a little x preceding the stats (xBA, xSLG, xwOBA, etc). Although this data is not magical, it helps us to predict the future with little error, even if it was not clear. it is designed increase the predictive value. What ZiPS uses is something designed to be as predictable as possible. I have talked about this a lot to both hitters and pitchers. The expected statistics used by ZiPS are called zStats; I’ll let you guess what the “z” stands for!

It is important to remember that these are not predictions in themselves. ZiPS certainly doesn’t just look at last year’s pitcher’s zSO and say, “Cool, brah, we’ll just go with that.” But the data includes how events unfolded, and is more stable for each player than actual statistics. That allows the model to shade the projection to one side or the other. And sometimes it is extremely important, as in the homers who were allowed to have jars. In neutral statistics, homers can be highly degraded, and pitchers’ home run ratings are more predictive of future homers than actual homers allowed. Also, if a pitcher “underachieves” or “overachieves” by a certain count, that’s when ZiPS believes actual performance is higher than expected.

One example of the latter point is Tyler Anderson. He has a track record of not doing well what ZiPS expects, so much so that ZiPS doesn’t really believe zStats at this point (more on Anderson below). Expected statistics give us useful information; they don’t mix magic.

What also pleases me is that zHR is surprisingly low in homers this year. There have been 2,076 home runs hit in 2024 as I write this, yet before making league-wide adjustments, zHR thinks there “should” be 2,375 home runs, a difference of 299. That’s a big difference; zHR has never dropped more than 150 home runs in the league all season, and knows that these home runs hit hard in April/May and the summer is still to come. That makes me wonder about the sudden drop in offense this year. It’s not a performance change, as I ran again in 2023 with the current model (with any training data from 2023 removed) and there were 5,822 zHR last year compared to the actual total of 5,868 homers.

Let’s start the jars with summary data.

Extreme zFIP (6/13, 150 min. TBF)

zFIP Underachievers (6/13, min. 150 TBF)

As you can see, ZiPS is not buying Trevor Williams as an ace. While he doesn’t get the benefit of an insanely low BABIP, he’s also allowed just two homers, with a 0.32 HR/9, which isn’t something anyone can sustain long-term. The Phillies have two starters among the overachievers, but that’s not necessarily bad news, as both Ranger Suárez and Christopher Sánchez have pretty solid zFIPs. ZiPS projects the Phillies to finish the season with the best rotation in the majors, one of Philly’s best rotations, and the best rotation of the Wild Card season. Taijuan Walker is the biggest forward here; he has the second-worst zFIP of any pitcher with 150 batters faced, at 5.50, better than only Michael Soroka at 5.51.

ZiPS thinks that Bryan Bello has been robbed a bit this year in terms of results, enough for me to pick him up in both of my favorite leagues. While zFIP was a little concerned about Dylan Cease, it helps to have a positive outlook on other San Diego pitchers like Joe Musgrove and Michael King.

I am including all the leaders here, by request.

zFIP Overall Leaders (6/13, 150 min. TBF)

What’s most interesting to me about this list is that zStats has some faith in breakout pitchers this year, who you might expect to overperform and be subject to significant regression in the definition. Although some of the latter are possible, there is more meat in the games. Detroit has two top starters here; Skubal’s emergence as one of the best hitters in the game is legit, and there’s strong evidence that Jack Flaherty’s revival is more than a fluke. A similar resurgence for Chris Sale seems real, and zStats supports breakouts for Tanner Houck, Jared Jones, and Cole Ragans, among others. I was really surprised by Luke Weaver’s performance this year; I think I was too early to write him off.

Extreme Winners of zHR (6/13)

Name HR HR Differences in zHR
Christopher Sánchez 1 6.0 -5.0
Logan Webb 4 8.5 -4.5
Luis L. Ortiz 1 5.5 -4.5
Kevin Gausman 8 12.1 -4.1
Jon Gray 3 7.0 -4.0
Trevor Williams 2 5.9 -3.9
Cole Irvin 6 9.6 -3.6
Burch Smith 1 4.6 -3.6
JP Sears 8 11.6 -3.6
Cole Ragans 4 7.4 -3.4
Sean Manaea 6 9.4 -3.4
Adrian Houser 3 6.4 -3.4
Adrian Morejon 0 3.2 -3.2
Albert Suárez 1 4.2 -3.2
Joe Mantiply 0 3.2 -3.2
Dylan Cease 9 12.1 -3.1
Mitch Keller 6 9.0 -3.0
Matt Strahm 0 3.0 -3.0
Kenley Jansen 0 3.0 -3.0
Tyler Anderson 10 13.0 -3.0

zHR Underachievers (6/13)

Looking at active pitchers since 2015, zStats has rated less than 11 pitchers with at least 10 total homers. Three of them – Lance Lynn, Kyle Gibson, and Steven Matz – have been signed by the Cardinals in recent years. I’m not sure what that actually means, but it’s at least worth paying attention to. Here you can see why zStats likes Hunter Brown so much; he’s actually hard to hit in the air and hard to hit hard, so there’s at least some reason to think his gopheritis this season might be out, or at least something he can fix as he develops as a hitter. Hendricks is an interesting story because he doesn’t really exist yet good, but enough here that more patience may be required. For someone who allows a ton of homers, he sure doesn’t hit much.

Pitcher home run charts are the most important of any zStats because, unlike most other numbers, zStats doesn’t just approximate the actual stats with the predictive value, they dominate them. HR/9 is just a terrible number for hitters and has led to many bad deals for many teams, and many great ones for the Dodgers! xFIP shouldn’t work as well as it does, it’s just that the home run stats are so bad that you’re actually better off, if given the choice, taking everyone else’s league average than one year’s HR/9 stats. And that’s a pretty ridiculous thing to do when you think about it.

zBB Underachievers (6/13)

zSO Underachievers (6/13)

Tyler Anderson’s year gets even weirder with these charts. His ERA (2.63) is currently more than two full runs below his FIP (4.72), but there’s more to the puzzle here if you dig into the numbers. He just missed the zFIP chart (ZiPS thinks his zFIP should be 0.23 runs better) but even more instead. Among active pitchers, Anderson is the pitcher who has used zSO the most – with 89 strikeouts – since 2015, a gap of more than 20 more strikeouts than the next guy (Wandy Peralta at -67). His career strikeout rate is absolutely terrible for a pitcher with an average league contact rate. The Dodgers are getting a great season from him in 2022, but they can’t even figure out how to stop his strikeouts.

As with hitters, strikeout rate is less important in modeling hitting, while contact numbers are incredibly important. And like hitters, decision metrics mean more when you’re walking, like first-strike percentage, which I’ve long used as a leading indicator of walk improvement/decline.

I’ll be using zStats again this season, in late August, and we’ll re-evaluate how zStats did against the actual numbers with two more months of data.

Source link

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button