Hitter xwOBA Overperformers — June 30, 2021

Yesterday, I listed and discussed the hitters with at least 200 PA who have most underperformed their xwOBA marks. Today, let’s now flip to the overperformers, those who have posted wOBA marks most above their xwOBA marks.

xwOBA Overperformers
Name BB% K% ISO HR/FB BABIP wOBA xwOBA wOBA-xwOBA
Joey Wendle 6.7% 20.9% 0.184 12.7% 0.335 0.347 0.294 0.053
Tucker Barnhart 9.4% 29.7% 0.152 11.8% 0.383 0.339 0.288 0.051
Nick Madrigal 5.1% 7.9% 0.120 5.6% 0.324 0.337 0.287 0.050
Jared Walsh 7.8% 29.0% 0.272 31.6% 0.353 0.378 0.330 0.048
Cedric Mullins II 9.0% 19.1% 0.221 14.1% 0.363 0.393 0.347 0.046
Trent Grisham 11.4% 24.6% 0.222 15.4% 0.341 0.369 0.323 0.046
Randy Arozarena 9.3% 27.3% 0.158 14.7% 0.346 0.334 0.289 0.045
Yuli Gurriel 10.7% 9.1% 0.185 9.9% 0.331 0.388 0.343 0.045

With Vidal Bruján still patiently waiting for his MLB callup, sitting atop the xwOBA overperformer list isn’t a good place to be for Joey Wendle. A career best HR/FB rate and ISO have pushed his wOBA up to a career best, but an xwOBA below .300 suggests the results have been far better than deserved. However, if you review his historical wOBA vs xwOBA marks, you will notice that he has typically outperformed his xwOBA marks. In fact, for his short career, he sports an acceptable .322 wOBA, but a measly .299 xwOBA. That kind of outperformance suggests Wendle might be doing something that is missed by the xwOBA calculation. That said, I would still caution that 1,364 PA isn’t a clearly large enough sample to make the determination he’s an inherent xwOBA outperformer. That’s basically two full seasons and I would expect we would need more to be sure. Still, it’s likely he’s better than xwOBA suggests, but his job will remain at risk if a slump hits and the Rays decide to recall Bruján.

Well duh, Tucker Barnhart’s .383 BABIP isn’t real. It’s actually a bit more surprising than you might think, as he sports an elite 29.3% LD% and tiny 2.9% IFFB%. But I guess his balls in play just aren’t of high enough quality regardless of the batted ball type. Statcast calculates his xBA at just .214, versus a .270 actual. It’s hard to find a catcher replacement so if you’re an owner, there’s likely not much you can do besides cross your fingers xwOBA doesn’t act like a crystal ball in Barnhart’s case.

What was so intriguing about Jared Walsh’s small sample breakout last year was the massive improvement in strikeout rate. It didn’t last. His strikeout rate has more than doubled and is back up to where we would have expected given his non-2020 history. He’s still hitting for tons of power, though Statcast thinks his SLG should be 100 points lower than it actually is. He also has no business hitting .283 and you could see all the pop-ups in his batted ball profile that you would think would make it difficult for him to post a .353 BABIP. I’m guessing most fantasy owners are still not fully buying into him yet, but if you do own him, it couldn’t hurt to see how high you might be able to sell him.

Cedric Mullins II has been one of fantasy’s most pleasant surprises, contributing strong four category production. Statcast calculates that his BABIP is in no way deserved, as its xBA is just .275 versus a .315 actual batting average. If that batting average does crater, it’s going to cut into his stolen base opportunities and take a big bite out of his value.

Trent Grisham has followed up well from last season’s breakout, but has traded some home run power (drop in HR/FB rate) for more hits on balls in play (higher BABIP). Statcast isn’t buying the jump in BABIP, which like Mullins, could cut into Grisham’s stolen base opportunities if it declines without a drop in strikeout rate too.

We were all eager to find out how Randy Arozarena would follow up his shocking 2020 season. He has continued to display a nice combination of power and speed, but he’s hitting far too many ground balls and a significantly lower rate of the flies he has hit have gone for a homer. Yet, Statcast still calculates an xwOBA below .300. It’s seemingly all about the BABIP, as it calculates an xBA of just .212 versus an actual .263 mark, likely thanks to a lowly 14.2% LD%, which makes it extremely difficult to sustain an inflated .346 BABIP. The power looks fine though and Statcast even calculates he may deserve an extra 0.8 homers. If he does hit a slump and his BABIP and batting average begin to tumble, he could quickly find himself getting dropped from the top of the order.

At age 37, Yuli Gurriel is having his best offensive season, driven partially by a double of his career walk rate. He has now actually walked more often than he has struck out. That’s great, but Statcast thinks his current .331 BABIP has been done using smoke and mirrors. His career best BABIP sits at just .308, but this year his FB% sits at the second highest rate of his career, while his LD% sits at the lowest mark. That’s usually a combination that would lead to a lower BABIP, not a significantly higher one. That said, while the batting average will almost certainly fall, the rest of his skills package remains rock solid and he bats in the middle of the lineup for the best offense in baseball.





Mike Podhorzer is the 2015 Fantasy Sports Writers Association Baseball Writer of the Year. He produces player projections using his own forecasting system and is the author of the eBook Projecting X 2.0: How to Forecast Baseball Player Performance, which teaches you how to project players yourself. His projections helped him win the inaugural 2013 Tout Wars mixed draft league. Follow Mike on Twitter @MikePodhorzer and contact him via email.

newest oldest most voted
Joe Wilkey
Member
Member
Joe Wilkey

Regarding Wendle, a few things. First, this again doesn’t account for plate discipline, I feel like Wendle should have fewer Ks than he has, but I beat that drum yesterday. Other than that, there are two other things that xwOBA on StatCast does not do a very good job accounting for: stadium and horizontal batted ball direction.

I have developed my own xwOBA based on StatCast numbers (*nerd alert*), and I have Wendle’s xwOBA as nearly identical to his actual wOBA. Some of that is due to the Ks, but there’s more there. If you look at Wendle’s FB on StatCast, his xwOBA is 141 points lower than his actual wOBA. Looking under the hood at his FB, however, I feel like that difference should be about half that.

For starters, if you look league-wide, the wOBA-xwOBA for pulled fly balls has never been less than .226 in the StatCast era, and that’s this year with the new ball and skewed toward cold weather games. Second, Yankee Stadium has a wOBA-xwOBA of .428 for lefty pulled fly balls in the StatCast era. Needless to say, pulling fly balls in Yankee Stadium is in the best interest of lefties. Wendle is a lefty who plays in the AL East, and while it’s not half of his games, he does get to play a lot of games at Yankee Stadium.

I’m going to examine the fly ball where my system differs most from StatCast: https://baseballsavant.mlb.com/sporty-videos?playId=35c8a17d-020a-4621-9fe5-1cc8443c07bd

It’s a fly ball to right field with an EV of 93.6 and an LA of 30. StatCast has the xwOBA on this as .317, but it went for a HR (a 2.039 wOBA). My system spits out an xwOBA of 1.370, a full 1.053 points higher. The difference is mostly Yankee Stadium. There have been 10 lefty pulled FB in Yankee Stadium over the StatCast era with EV between 93 and 95 and LA between 29 and 31, and 8 of them have gone for HR, with the other two going for outs. This yields a 1.619 wOBA. For this combo of EV/LA to the lefty pull field, the rest of the stadiums in the league have a combined 1.076 wOBA, so it’s not just Yankee Stadium, either.

Now, we can argue how much control batters have over batted ball direction and whether they have certain approaches that change based on stadium, but xwOBA is itself a descriptive metric, so I believe it should do a better job taking into account even basic batted ball direction (pull/straight/oppo) and stadium effects.

tonycpsu
Member
Member
tonycpsu

Your ideas are intriguing to me, and I wish to subscribe to your newsletter. Do you have your results published anywhere?

What does your model say for Mullins? Camden doesn’t have the Yankee Stadium short porch, but it certainly is a good place to pick up a couple of wall scrapers to pad the HR stats, and he does play a good number of games in the Bronx as well.

Joe Wilkey
Member
Member
Joe Wilkey

So I actually just put them in a useable format here: https://docs.google.com/spreadsheets/d/1p87-r6s26iBGoqp0nD12th-cXLdOZHo0E3J3I7FTeNQ/edit?usp=sharing

I don’t update it all that often, since I have a full time job and kids, but I usually do get to it once a week at least. I use StatCast data for all of it, I’ve actually found that I can model K% for both hitters and pitchers based on about 22% of the pitches they see (on average) and BB% based on about 53% of the pitches they see (it’s a little trickier, but it’s still not bad). My batted ball data attempts to account for stadiums and shifting, I’m still working on incorporating speed, but I’m not sure it’s as important as everyone thinks.

As for Mullins, I do think he’s actually a little light on his home run total, but they’ve generally turned into doubles. Generally, I do think he’s outperforming somewhat, but more to the tune of 20 points of wOBA rather than 46.

I do think the ball has had an effect on the HR totals this year in general, so it’s best to compare one player’s xStats to everyone else’s xStats, since I’m basing all of this on historical data in the stadiums, which does not account for the new ball.

Anon
Member
Anon

There are a number of errors in xwOBA and I’ve been assuming for awhile that there will be an adjustment to it eventually. What’s astounding is the adjustments are so obvious. In addition to stadium and batted ball direction there is speed. Of course Miguel Cabrera is going to fall short of his xwOBA year after year – dude is so slow that the infield gets to play him 15 ft off the infield dirt

Joe Wilkey
Member
Member
Joe Wilkey

Cabrera’s xwOBA difference is almost exclusively due to his stadium and not his speed. He has actually exceeded his xwOBA on GB in the StatCast era, and his xBA on line drives is roughly equal to his BA, so if the infielders are playing back, they’re not catching that many extra LD. His xwOBA on FB is drastically higher than his actual (.564 v. .429), the difference at home is mammoth at .203 (.615 v. .412) and the difference to straightaway at home is comical at .531 (.787 xwOBA v. .256 wOBA). Center field in Comerica is where fly balls go to die, I was so pumped as a general baseball fan to see Nick Castellanos depart that offense killing outfield.

I really don’t think speed has that big of an effect on xStats, honestly. There’s maybe a handful of GB in a player’s season where he’s able to leg out a grounder (or the other way around) and there might be an extra base taken here and there on a line drive, but I’m not sure it has the massive effect people seem to think it does.

Anon
Member
Anon

I honestly had never looked that up so I’m surprised on Cabrera. But I still think speed is a factor. I did a search for players who have had 500 GB in the Statcast era and it’s hard not to be struck by all the fast guys that have massively outperformed xwOBA (DDS, Odubel, Dee Gordon, Hamilton) and all the slow guys (especially slow lefties like Justin Smoak) who have massively underperformed. I’m sure the shift is a huge factor but there are plenty of slow righties who have underperformed as well (EE, Sal Perez, Cron) .

Intuitively it makes sense – Statcast sees all 95 mph grounders as the same and assigns everyone the same xwOBA on that but the infield is having to play Dee Gordon in and Miggy and all the other slow hitters back. The fielders just have a lot more time to make a play on the grounder from the slow guys than they do on the grounders from the fast guys. There’s no way that doesn’t suppress wOBA for the slow guys.

Joe Wilkey
Member
Member
Joe Wilkey

There’s probably some signal there, but it’s still more to do with the shift than with speed. If you look at all players who have at least 500 GB with no shift and compare it to their sprint speed (N=153 over the statcast era), the r^2 for wOBA-xwOBA is .074, or almost nothing. Using a straight linear regression between sprint speed and wOBA-xwOBA, 1 ft/s of sprint speed is worth .004 points of wOBA. The largest number of GB for a player in a given year is about 250, and someone like Buxton is about 3 ft/s faster than league average, which calculates out to an additional 3 singles over the course of a year on the extreme end of things.

Furthermore, I’m pretty sure xwOBA buckets all GB together, so the xwOBA for GB that are not shifted is reduced by the shifted GB, and faster players tend not to be shifted. The overperforming of xwOBA due to speed may not be due to positioning explicitly, but due to the xStats for a player who is rarely shifted (like a Tapia) being artificially reduced by the pool of players who are shifted.

Joe Wilkey
Member
Member
Joe Wilkey

Please note that .004 of wOBA for 1ft/s is ONLY for GB, so when you lump it in to the entirety of a player’s profile, if you assume league average K%/BB%/GB%, it’s .001 of wOBA over a season, which is basically noise.