Gerrit Cole’s Bummer Summer

The last time we saw Gerrit Cole in an Astros uniform, he wasn’t actually in an Astros uniform. He was, instead, in a Boras Corporation cap, ready to chart his own course through the league after a dominant run in Houston. When he signed with the Yankees, it felt almost preordained — one of the bright stars of baseball, either the best pitcher in the league or a close second, on the most storied franchise in the game. We get it — great players like the Yankees, and the Yankees like great players.

One look at the surface-level statistics will tell you that something hasn’t panned out in 2020. A 3.63 ERA? A 4.69 FIP? Thirteen home runs allowed in only nine starts?! He’s allowed a home run in each start, which is about as disastrous as it sounds. Heck, even his record tells you something is up; he’s 4-3 this year on an underachieving Yankees team, and while wins and losses are silly contextual statistics, Cole went 35-10 the last two years. Something is clearly up.

Far less clear? What that “something” is. There are some easy ways pitchers fail, ones you can see from a mile away. They lose velocity, and their fastballs become newly hittable. That hasn’t happened to Cole, though, at least not really:

Gerrit Cole, Pitch Velocity (mph)
Year FB SL CU
2015 96.5 87.7 82.1
2016 96.0 88.3 81.8
2017 96.3 88.5 80.8
2018 97.0 89.1 82.9
2019 97.4 89.5 82.8
2020 97.0 89.1 83.8

Starters can also lose feel for one of their pitches, and change their pitch mix to compensate. That hasn’t happened either:

Gerrit Cole, Pitch Usage
Year FB SL CU
2015 50.9% 21.4% 7.8%
2016 50.1% 17.8% 9.9%
2017 41.8% 17.3% 12.2%
2018 53.4% 19.9% 19.3%
2019 53.6% 23.1% 15.5%
2020 53.5% 24.8% 16.5%
Note: FB is four-seam fastball only

Uh… maybe he’s the victim of a poor early-count approach. He’s throwing fewer fastballs this year to start batters off, but just as many pitches in the zone. He’s not doing it by throwing more curveballs and sliders in the zone, either:

Pitch Usage on 0-0
Year Fastball% Zone% Zone Brk%
2015 71.2% 52.4% 43.6%
2016 73.4% 53.3% 46.0%
2017 64.4% 55.7% 58.2%
2018 61.7% 57.8% 51.0%
2019 57.0% 56.4% 52.8%
2020 54.5% 57.8% 50.0%

In other words, Cole is throwing fastballs less often to start, but he’s making up for it by throwing them in the strike zone more often. Sounds dangerous. Are batters suddenly teeing off on him on 0-0? Nope! They’re actually swinging less than ever, and the whole thing is too small-sample to matter anyway. He’s getting to 0-1 54.5% of the time, in line with his dominant 2019. Next!

Okay, if it’s not about his behavior to set the tone, what about his behavior when there’s a chance to end things? Maybe Cole has lost his way with two strikes, and is simply giving batters too many bites at the apple. In today’s game, you don’t want to be in the business of extra-apple-bite-giving.

There’s only one inconvenient truth about this narrative: it’s not true. When Cole throws pitches with two strikes, he’s still in attack mode. His put away rate, the rate at which two-strike pitches turn into strikeouts, has been stable. He’s going to his slider more often, and most metrics have that as his second-best pitch, behind only his fastball. All in all, his behavior looks downright logical:

Two-Strike Behavior
Year Put away% Slider% SL Put away%
2015 20.4% 27.7% 27.4%
2016 16.8% 23.3% 22.8%
2017 19.5% 23.1% 21.0%
2018 25.2% 26.6% 24.7%
2019 28.6% 28.7% 27.4%
2020 24.4% 34.8% 25.0%

If you’re looking for a sliver of a change, Cole’s slider is ever so slightly less devastating than it was a year ago. That’s roughly two sliders’ worth of difference, though; hardly worth mentioning. Batters are spoiling the pitch slightly more often, but that makes sense given that he’s throwing it more. Nothing looks amiss here.

No, if you want to find worrisome trends in Cole’s 2020 season, you need to look all the way down to contact management. That phrase often merits a derisive laugh: “Oh, that guy? He’s only getting by on contact management. Can’t miss a bat to save his life.” That’s clearly not Cole. But all the same, those home runs are the problem, not some esoteric pitch mix question. Opponents are simply hitting the ball harder when they connect:

Batted Ball Metrics
Year Avg EV Barrel% Hard Hit%
2015 88.3 3.8% 35.1%
2016 87.8 2.7% 30.9%
2017 86.3 7.4% 33.5%
2018 89.1 6.2% 39.8%
2019 87.6 5.7% 35.5%
2020 91.1 10.2% 47.2%

Worst of the Statcast era in every category? It’s not great! Average exit velocity can deceive, because the distribution matters, but 3.5 miles an hour is a lot. Barrel rate and hard hit rate don’t suffer from those distribution issues, and they’re telling a scary story. His average launch angle against has increased, as well. That’s a horrible stat — a pop fly and a grounder average out to a line drive. But it still has qualitative use, because it tells us that Cole’s opponents are putting the ball in the air more frequently, and we already know they’re hitting the ball harder.

Alex Chamberlain has done some great research into explaining what pitchers can and can’t control. His finding — not all that surprising when you think about it for a minute — was that pitchers have more control over launch angle (largely by where they locate their pitches) while batters are in the driver’s seat when it comes to exit velocity. Sounds like good news — Cole’s hard hits aren’t his fault!

Yeah, uh, about that… Cole scored terribly last year in Chamberlain’s expected launch angle metric. He was in the bottom 15 pitchers in baseball when it came to the expected percentage of balls hit at dangerous angles — between 12 and 27 degrees. One silver lining? He fared much better when it came to expected percentage of balls allowed between 18 and 22 degrees, the sweetest of sweet spots, which means he was throwing to locations that gave batters a little less margin for error.

That’s no death knell for pitchers — there are plenty of good pitchers high on this list, and they get by on swing-and-miss stuff, which Cole has in spades. Jacob deGrom, for example, ranks similarly, and I mean, he’s Jacob deGrom (groundbreaking analysis here). Ignore where he threw his pitches (that’s what Alex used to come up with expected rates) and focus instead on what batters did, and Cole was excellent in 2019. He had a launch angle standard deviation of 29.5 degrees, near the top of the league (deGrom, for comparison, checked in at 29.8). That number is a little worse this year, at 27.5 degrees. He’s still doing a decent job of keeping batters from repeating line drives all over the place — though due to a change in tracking technology, it’s probably best not to read too much into it.

Furthermore, Cole allowed 33% of batted balls to be hit in the sweet spot (between 8 and 32 degrees) last year. He’s at 35% this year (league average 35%) — again, the tracking has changed — and yet, those darn home runs keep on coming. Throw math at it all you want — batters keep hitting balls right out of the park.

Rather than look at any top-level metrics, then, let’s look at the home runs. Eight have been on fastballs, two on sliders, two on curveballs, and one was on a changeup. That’s a roughly average distribution for him — fastballs are the culprit on a majority of his home run balls. There’s nothing obviously wrong with those eight fastballs either: they’ve averaged 97.1 mph, faster than his average fastball this year. If there’s anything truly scary about them, it’s the location — six have been over the heart of the plate. This pitch isn’t exactly putting Mike Zunino to the test:

And Adam Eaton got a gift here with two strikes:

In fact, two-strike homers were a problem for Cole in 2019 as well, when he gave up 14 of them. That’s partially because he attacks the zone more than league average with two strikes, putting hitters in a bind: take, and you might be out anyway, but swinging at one of Cole’s pitches isn’t usually a great outcome either. But wait! He’s throwing in the zone on two-strike counts at the lowest rate of his career this year and still giving up these dingers. Ugh!

At this point, I’m grasping at straws. Interestingly, though I think we might have found the answer up above, before we got into any of this spray management or launch angle this-and-that. You see, Cole has never particularly excelled at limiting damage when batters make contact. He’s allowed a .395 xwOBA (.375 wOBA) on fastball contact since 2018, with the league nearby at .398.

He’s been a little better at avoiding loud contact with his slider, a little worse with his curveball; for the most part, though, he’s just a pitcher when the ball gets put into play. Remember Chamberlain’s findings, that pitchers control launch angle? Well, Cole’s nothing special there. How surprising is it that he’s not the king of weak contact?

Cole’s genius, then, lies in missing bats. In Houston, he missed bats with everything — batters swung and missed at 19.3% of the fastballs they saw from Cole last year, which is an absolutely outrageous mark. No other starter has finished within three percentage points of that mark in the past five years. That’s Josh Hader territory, only as a starter with two plus secondaries. This year, that number has plummeted to 12.7% — still the 12th-best mark in baseball, but a far cry from last year.

I’m skeptical that Cole will keep giving up homers at such a preposterous rate. But I’m also skeptical that he’ll ever regain his 2019 form, if only because it was so far removed from anything we’ve seen from other starting pitchers. A slight drop in his level doesn’t make him average; it merely makes him less-than-Cole. His fastball has a half-inch less rise on average this year, a half-inch more arm-side run. His average velocity is off a half a tick. Small variations, but variations nonetheless.

Maybe Chamberlain’s research is right, and Cole’s pitching style is vulnerable to balls hit at the worst possible angles. Maybe Cole’s past results are the better guide, and he’ll continue doing a decent job of inducing a wide array of launch angles (that standard deviation number from above). Most likely, though, batters will continue to do roughly average damage when they connect. That’s a good guess for all pitchers, and Cole hasn’t done anything in particular to make me doubt it for him. What’s wrong with Gerrit Cole? I don’t think it’s the home runs. I think it’s the whiffs — and it’s not really wrong, just less right than it was last year, ridiculous home run streak aside.





Ben is a writer at FanGraphs. He can be found on Twitter @_Ben_Clemens.

27 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
RonnieDobbs
3 years ago

Its all sign stealing related.

RonnieDobbs
3 years ago
Reply to  RonnieDobbs

Alternate theory – batted ball data provides very little insight to understanding anything. HR allowed are one of the worst metrics that you can use to gauge pitcher performance, but it ends up being one of the most commonly used. Need proof? Look at literally anyone who is both good and has pitched many full seasons. What is normal for Cole? I imagine that he was pitching over his head last year which is probably half of it.

tz
3 years ago
Reply to  RonnieDobbs

HR by itself won’t tell the story. But this year Cole has allowed a career high % of fly balls, hard hit balls, and balls being pulled. That might indicate that hitters have caught up with the post-Pittsburgh version of Cole and do a better job of jumping on his four-seamers in the heart of the zone (like Juan Soto did in last year’s WS). Or maybe he’s tipping his pitches somehow.

Or it could be just crazy random noise over 52 IP. If Jason Vargas can ride random noise to an All-Star appearance (which he did in 2017), Gerrit Cole can have random bad luck in equal measure over a couple of months.

awy
3 years ago
Reply to  tz

i think this is on the right track drawing out how across the board cole’s advantage on the fastball has eroded from the stats. hitters are having a lot more success squaring up to his fastball, which has yielded xSLG of over .550 this year vs .321 in 2019. he’s also inducing roughly half as many popups as last year, and the % of his 4seamers being put into play has also nearly doubled.

at a high level of abstraction, this is regression from one of the most dominant fastball seasons ever, so in that sense it’s not a surprise. but still interesting to think about exactly what’s changed in baseball terms to cause the decline in fastball dominance.

of particular interest to current state of pitching analytics is that cole’s fastball’s movement and spin etc are not so different. thus, this is a situation where the under-quantified areas of pitching could be playing a big effect and remind us how much we still do not know about pitching from current best available public stats.

there’s tunneling, he’s using the changeup a bit less, from 7.5% to 5%. but we run into data sparsity here with only 6 batted ball events from the changeup so far. it’s possible that a lot of the difference is from lack of tunneling effectiveness, particularly seeing how hitters are having a lot more success hitting his fastballs down in the zone where his changeups usually end up. part of the tunneling dynamic is effect from shifting hitters’ approach. if hitters are deciding that they are fine hunting for low zone fastballs and elevating them, then the lack of changeup threat can play a very large role here.

then, there’s deception. cole’s never had a reputation as a particularly deceptive guy, but it’s something he’s said to be working on. he specifically cited adding deception as a reason for tweaking his delivery recently, so maybe he was feeling it was a problem. with the aforementioned ways cole’s fastball has declined in effectiveness, it’s really the seeming collinearity of these declines that seem to indicate a common cause, hitters are less overwhelmed by the pitch. deception can fit the bill, but i don’t know why the effect seems to be limited to fastball only, while his breaking pitches are a bit more effective results wise. (probably tunneling?)

the effect of deception upon pitching performance seems to be large but fragile, with few pitchers able to sustain prolonged periods of it. those who do are often in elite company. maybe some variation of deception is causing a large shift in how hitters fare vs cole’s fastball, and it’s a difficult thing to replicate.

tunneling or deception, it takes a lot more playing baseball knowledge to divine than simple statistics. as data reveals more dimensions of the problem, specific knowledge about the game becomes more important for the baseball analyst.

awy
3 years ago
Reply to  awy

btw i neglected to look at the curveball situation, which he’s apparently throwing way lower and more out of zone this year. that could affect tunneling too.

RonnieDobbs
3 years ago
Reply to  awy

I doubt he is intentionally throwing it worse. That is good insight though. I am not sure why you have a downvote – maybe it was meant for me.

RonnieDobbs
3 years ago
Reply to  awy

I like how we don’t really acknowledge that three could have been luck in the past. In my opinion, the idea that there has been any execution or positive league wide change in 2020 is absurd. They don’t even practice. I have seen so many events this year where nobody has any idea what to do. The idea that hitters are adjusting is laughable to me. Maybe Cole is not in shape for the season? There are so many factors that could contribute to anything. Data and buzzwords will never solve the problems which you are discussing. You have blind faith in metrics and progressive baseball ideals that will never be rewarded.

duckybuckinfent
3 years ago
Reply to  tz

Lotta good stuff in the article *and* the comments.
I’m new here. Hoping I can hold my own with you fellas, serious.

Here’s what I’m wondering about.
All those little things: a little less velocity, a little less rise, a little less spin, a little lower arm slot, etc.
Could be just simply age stuff, man. (i.e. the very outer edge of: Decline.

Thankfully – as a Yankee fan – he’s got such big stuff and he seems to really embrace modern pitching philosophy/analytics that it seems like he’s a guy who’ll age well.

RonnieDobbs
3 years ago
Reply to  tz

A lot of it is just luck – both bad this year and good last year. I am certain that the quality of baseball everywhere is through the floor this year. I doubt anyone caught up with anything. It is more likely that Cole is off than these essentially 40-man September rosters are on.

tz
3 years ago
Reply to  RonnieDobbs

Yep, a lot of anything going on this season has to be taken with a grain of salt, maybe a whole salt lick worth. That said, if anyone’s going to get a little “caught up with” this year, it might be a star pitcher facing the same teams over and over due to the unbalanced schedule. The Rays faced Cole three times in less than a month, and hit him pretty hard in the last of those starts.

And besides, what better “man bites dog” story than a guy like Cole getting taken deep by the likes of Mike Zunino? Heck, David Schoenfield posted a virtually identical article today on ESPN.com. 🙂

awy
3 years ago
Reply to  RonnieDobbs

“hr allowed” isn’t even what ‘batted ball data’ looks like nowadays. as for it being talked about, besides homers being influential for run prevention, it’s also a component of FIP. people want to try to figure out which factors affecting the homer result is noise and which ones are skill/meaningful.

RonnieDobbs
3 years ago
Reply to  awy

Actually HR allowed is what people worship the most. FIP largely feeds off of it as do all metrics that weight pitcher value. There has been no progress in pitching analysis over the Statcast era. Places like this still advocate for ERA and Win leaders more often than not. We have low-end middle relievers pitching more innings and fewer elite SP innings. It is pretty clear to me that nobody is increasing their understanding of pitching through batted ball data. Statcast data helps people who understand very little about baseball to think they understand more than they do. Don’t be shocked when nobody ever solves any of these problems. I think I know what you are trying to say, but you are still a few steps behind. HR allowed are largely lucky – batted ball data doesn’t shed any light on that. A good swing is a good swing in a gamefeed, broadcast and on the field. it is not representative of anything more than what it was. Batted ball metrics are a generation or two behind a basic understanding of how on-field baseball works. I don’t see how it would ever get ahead.. although I do enjoy entertaining the idea. When Cole’s HR luck turns around it won’t be by some progressive nonsense – that is the way things have always worked, but that won’t stop people from overlaying Statcast data over the real results and pretending that they had something to do with the result. In the end, this is a parasitic cycle of pretending that people off the field are sharing in Cole’s success and condemning his failures as a fixable problem.