Dylan Lee: (An unusual) Death to Lefties

Perhaps no other bullpen arm epitomizes the 2022 Atlanta Braves relief corps as well as Dylan Lee. The 28-year-old southpaw from the tiny town of Dinuba, California was drafted by the Marlins in the 10th round of the 2016 amateur draft, but was released by the Fish ahead of the 2021 season, options and all. The Braves snapped him up and gave him a cup of coffee late in the year, as well as making him the answer to the trivia question: who was the first pitcher to make his first Major League start in a World Series game?

Fast-forward to 2022, though, and Lee has been anything but a curiosity relegated to the trivia annals. Rather, he’s up to 1.1 fWAR on the year, with a 52/69/79 line and an xERA somewhere between his ERA and FIP. He has more fWAR than high-priced relief acquisition Kenley Jansen; he’s tied for 40th in fWAR among all MLB relievers. Not bad for a guy that was outright released rather than having his option years run down, right? Not shabby for a guy who spent the first six weeks of this season pitching at Triple-A Gwinnett (although that was partly specifically because he had options remaining).

That low-risk, high-reward performance is partly why Lee epitomizes this bullpen so well. The two arms ahead of him in fWAR are A.J. Minter (arbitration-eligible, earning under $3 million in 2022) and Collin McHugh (earning $5 million this season and next season). Not ahead of him are the aforementioned Jansen ($16 million) and Will Smith ($13 million), the latter of whom was jettisoned off the roster at the Trade Deadline given his horrendous performance. The other reason he’s the bullpen in microcosm? Despite his superior performance, the timing of it has been mismatched to some extent. On the season, the Braves are tied with the Astros for MLB’s second-best bullpen. However, said bullpen is only eighth in WPA. In other words, they’ve been great, but comparatively less great in the most important situations:

• Braves bullpen, low leverage: 3.38 FIP, 3.91 xFIP
• Braves bullpen, medium leverage: 2.54 FIP, 3.10 xFIP
• Braves bullpen, high leverage: 3.81 FIP, 3.85 xFIP

Lee has this same hallmark right now. Despite 1.1 fWAR, he has a WPA of, well, 0.00. He has 12 shutdowns to eight meltdowns, which is a pretty crappy rate. (The league, as a whole, has a rate of about 1.7 shutdowns to meltdowns, and that includes a pantsload of relievers way worse than what Lee’s done this year.) While Lee’s xFIP is basically the same across the three leverage categories (3.10, 3.18, 3.19), he’s had the unfortunate affliction of giving up his longballs in higher-leverage situations, leading to an FIP of 1.61 in low leverage, but 3.98 in medium leverage, and 4.31 in high leverage. This homer incidence isn’t really Lee’s fault, though, it’s just the sort of thing that happens when you slice up small-sample events into even smaller disparate chunks.

As the Atlanta relief corps helped deliver a giant sweep of the Mets, Lee was smack-dab in the middle of the action. The Braves used nine pitchers in the series, and Lee finished second in WPA with 0.20, just barely behind Raisel Iglesias (0.22), who had the advantage of pitching in all three games. Lee actually led the entire Atlanta pitching slate in WPA in both Saturday’s and Sunday’s games.

---

I’ve now spent 600 words just on Dylan Lee, the 2022 relief guy, and haven’t gotten to what I really want to talk about: Dylan Lee is Death to Lefties this year, but the way he’s doing it is weird. Let’s start with some basics.

For the 2022 season...

• League-average: .309 xwOBA, 3.97 FIP, 3.97 xFIP
• Relievers: .302 xwOBA, 3.87 FIP, 3.95 xFIP
• Lefty relievers: .304 xwOBA, 3.94 FIP, 3.99 xFIP
• Lefty relievers, against lefty batters: .282 xwOBA, 3.45 FIP, 3.72 xFIP

Essentially, relievers are slightly better than starters this year, but lefty relievers are slightly worse on the whole, even though they do a good job of stifling lefty batters. Now, let’s do Lee:

• Overall: .252 xwOBA, 2.70 FIP, 3.14 xFIP (way better than your average reliever, though we already knew this)
• Against lefty batters: .205 xwOBA, 1.59 FIP, 2.45 xFIP (these are wow-type numbers — Lee has been against lefties something like what A.J. Minter has been against everyone this year)

Here’s where it gets a little weird, though. If I told you there was a reliever, and he was death to same-handed hitters, you would figure he has a fantastic slider, right? Based on a bunch of work from Max Marchi (2010) and Jared Cross (yes, the Steamer guy, 2015, see here: https://tht.fangraphs.com/forecasting-pitcher-platoon-splits/), we have the data that supports this. Lefty sliders, in fact, have some of the biggest “yes this will result in a big platoon split” influence out there. We can use actual data from 2022 and come up with something pretty similar.

You can see that some pitches are fairly low-split, namely four-seamers and changeups. Curves and cutters from right-handers also don’t have much of a split. Two-seamers and sliders have fairly meaty platoon splits, as do cutters from lefties. (Splitters are not pictured because they are so rare, especially from left-handers.) Fundamentally, if you are a lefty pitcher that eats lefty batters alive, chances are, you are throwing something in the cutter/slider/curve family that breaks away from the batter... right? And you definitely aren’t forcing a four-seamer in there, yeah?

Except, in Dylan Lee’s case, those things would be quite wrong. Because, as you will see below, Lee has got some stuff going on.

Lee’s slider is pretty much a lefty-lefty slider, in the xwOBA sense. It has a smaller platoon split. His fastball is better than the usual lefty-righty four-seamer in terms of xwOBA... but oh man, has it just eviscerated same-handed batters.

The exact reasons for this are hard to pinpoint. Lee’s slider-thing is more straightforward: it’s pretty cutter-y and is more of a one-plane pitch than something that really breaks away from lefty bats. His fastball has good shape, but lacks anything particularly strange about it that gives it an innate platoon advantage:

When I look at the above, what jumps out at me isn’t really the shape (though boy that slider is really getting into cutter territory) of anything, but the usage. Lee is about 50-50 with his pitches overall (setting the rarely-used, poor-shape changeup aside for a moment); against lefties, it’s more 40-60 fastball-slider. So, while a lefty facing Lee sees about 0.7 fastballs for every slider, in general, lefty-lefty matchups in 2022 have featured about 1.2 fastballs for every slider. (Even when adjusting for reliever pitch usage, it’s like 1.1, but generally not flipped.)

This style of backwards pitching might explain some of the fastball’s success against lefties. In other words, the fastball is the “gotcha,” the surprise out pitch after the hitter has been worn down by a steady diet of sliders. You can see this in Lee’s weird pitch type-by-count chart.

Basically, against a lefty, there’s a good chance Lee is throwing a slider, something that only tends to change on 2-2 and 3-2 (and also 2-1 and 3-1, though weirdly not on 2-0 or 1-0). The “traditional” method of pitching, especially with such an effective pitch as a lefty-lefty slider, is to get ahead and then put guys away with the hard-to-contact breaking pitch. But Lee is getting ahead with the slider, throwing more sliders, and saving the four-seamer for two-strike situations. That’s gotta be confusing for lefties, though it should be noted that there are plenty of lefty starters and relievers that also throw a ton of two-strike four-seamers, from lefty fastball champ Carlos Rodon to A.J. Minter.

There’s more, though. You might think from the above that given the “surprise!” usage of the four-seamer to lefties, its shape, and the fact that Lee is very good at taking advantage of that shape by throwing it right at the top of the zone or above, that Lee’s four-seamer is a master put-away pitch. But, it really isn’t, at least not in the conventional sense.

Against lefties, his four-seamer has a whiff rate of “only” 24 percent, and a put-away rate of “only” 22 percent. By comparison, his slider has whiff rates north of 40, and approaching 50 percent, against batters from both sides of the plate, with a higher put-away rate. Instead, what seems to be happening is that Lee is somehow benefiting from some horrid contact off a four-seamer that hitters probably shouldn’t be swinging at anyway.

Using the Statcast definition of the strike zone, the league swings at about 39 percent of pitches not in the zone, and connects on about 46 percent of those swings. Lefties are chasing both his fastball and slider at the same-ish 35-40 percent rate, but they generally get the fastball: 72 percent of out-of-zone fastballs that Lee has thrown to lefties this year end in contact. But, that contact is kind of the equivalent of “the dog that caught the car.” Congrats, you chased his four-seamer, here’s your spray chart:

There’s a bunch of fouls, some whiffs, and this piddly contact. Even if you relax the requirement that the pitches must be out of the zone, things don’t change much.

The crazy thing here is that Lee’s four-seamer, again, has good “rise,” and he’s throwing it almost exclusively up in the zone, yet getting a bunch of grounders anyway. This is a pretty generic example of Lee throwing a two-strike fastball that seems actively hard for a batter to beat into the ground, yet it happens anyway.

And I don’t even know what Kyle Schwarber is doing here, it’s like hitters are leveling out their swings to hit something too high because they know they can’t catch up to it, even though they probably can, because he’s throwing like 92 mph:

---

I do want to add one more note in here, because it’s an important one, and it’s not quite about lefties. There was a point, all the way up until around mid-August, where Lee lacked enough feel for his slider to really execute it consistently. His fastball was still great for the reasons listed above (.229 xwOBA-against through August 18), but his slider suffered from too much hanging and general plate-ness and had a very generic, bad-for-a-slider .311 xwOBA against.

Since that point, Lee has done a much better job spotting his slider, at the expense of fastball command. Given that we’ve only had this “switch” for about six weeks and a handful of appearances, it remains to be seen whether he can put everything together or not. But, even if this is best version of Dylan Lee we’re gonna get, it’s been quite a ride. He’s gone from afterthought to more-than-solid reliever, and a guy who torments same-handed batters with an unconventional, essentially backwards approach. Will it last? He’s a reliever. But it’s been head-scratchingly fun to watch so far.