clock menu more-arrow no yes

Filed under:

Analyzing similarities between Mitch Keller and Tyler Glasnow

New, 17 comments

Might Keller develop along a similar path?

MLB: San Francisco Giants at Pittsburgh Pirates Mark Alberti-USA TODAY Sports

Thus far in 2021, Pirates’ starting pitcher Mitch Keller has produced mostly less-than-stellar performances on the mound. This outcome — not a new one — has led some fans to believe it’s time to move on from the struggling Keller. Stuck in the ether of being too good for Triple-A baseball and too uncomfortable for Major League Baseball, the hourglass of Keller’s Pittsburgh tenure might be reaching its limit.

But with all this talk of Keller and poor performances, I’m reminded of a past Pirate pitcher who experienced his own highs and lows: Tyler Glasnow. The tall, Californian right-hander appeared in 56 games for the Pirates over three years, making 17 starts along the way. Now in Tampa Bay, Glasnow has been much improved, making 43 starts for a Rays team that dipped its toes into World Series waters in 2020.

With Glasnow’s and Keller’s expectations being somewhat similar in the black and gold, I wondered how much the two actually had in common. For some measurements, I was surprised to find similar metrics; for others, one had a clear advantage over the other. All in all, I was trying to determine whether or not Keller is to experience a similar fate as Glasnow: Struggle in Pittsburgh, only to find that special something elsewhere.

When watching both pitchers — though it was perhaps more prevalent with Glasnow — there’s something there. Whether that be lively fastballs or big breaking balls, it was/is hard to shake the feeling that these two guys were going to figure it out at some point. I am reminded of an early season start Glasnow made against the Tigers in Detroit in 2018. It was a plate appearance in which he walked famed slugger Miguel Cabrera, but Cabrera, once he got to first, marveled at the big bending nature of Glasnow’s curveball.

Similarly, Keller, though not necessarily to the same extent, has given up glimpses of what we all thought he could become. Present for Keller’s debut in Cincinnati, I was ecstatic to see the first start of the Pirates’ next ace. That outing, it turned out, would have mixed results. Ultimately responsible for six first-inning runs, Keller gave up hit after hit — including a Jose Iglesias home run — but went on to give the Bucs four innings, while striking out seven. Later that year, he’d face those same Reds in Pittsburgh, managing six innings while striking out nine and allowing only one run.

Keller has only started 23 games at the Major League level. When reminded of how short his stint has actually been — spread out over three years, including a shortened season — some fans find that to be a silver lining, hoping that he’ll still figure it out, now in his age 25 season; others think that he’s had plenty of time to more consistently prove his worth at the highest level of baseball.


Keller’s first season in MLB in 2019 saw an array of numbers competing for your attention. To start, his classical metrics indicated that he was bad. For example, Keller carried a 7.13 ERA to the finish line over 48 innings pitched. His WHIP, 1.83, didn’t lend itself to believing Keller was due for much success either.

But his SIERA was a very respectable 3.78, especially for his first season of Major League Baseball. More predictive metrics, like xFIP (3.47), also gave rise to the notion that Keller had suffered from a heavy dose of bad luck and that things should be trending upward in the young pitcher’s future.

Despite his proclivity to allow base hits — hitters had a .475 BABIP (.343 batting average) — he was striking out guys on a regular basis (28.6 K%) and didn’t walk batters at a terribly high clip (seven percent walk rate), netting a 21.6 K/BB ratio. Had Keller qualified by innings pitched, his K/BB ratio would’ve been 12th-best in all of baseball among starters, while his K% would’ve been 15th, and his BB% would’ve been tied for 31st in baseball.

Venturing into the 2020 season, hopes were there for Keller. He was due to get his first full season taste of Major League Baseball and, though the results from 2019 weren’t what he or the team wanted, much of the peripherals suggested he should expect an upswing in real production.

During a season in which everybody’s natural flow was altered, Keller again saw mixed results. Over his five starts, his real production had improved, just as had been hypothesized, but his expected production plummeted — a foreboding sign of what was to come.

Keller amassed a quality 2.91 ERA, which is in stark contrast to his 2019 output. But his FIP (6.75), SIERA (6.94), and xFIP (6.57) all skyrocketed. Those numbers are due in some part to his shrinking strikeout percentage and ballooning walk rate. The former fell to 18.4 percent, while the latter climbed to a staggering 20.7 percent. The reason for why he still experienced success, however, is because of actual hits earned against Keller. Teams were batting .132 against him, while his BABIP was an extremely low .104, which suggests he was the beneficiary of massive amounts of luck.

Certain websites and specific people (me) warned overly optimistic Pirates’ fans about the impending collapse foretold by more advanced statistics. Now in 2021, that’s precisely what we’ve gotten from Keller. He’s been inconsistent throughout the season and has largely failed to capture any sustained amount of success.

His ERA is back up to 7.81, while numbers like his FIP (4.96), SIERA (5.20), and xFIP (5.21) don’t tell the same optimistic story they once did. His WHIP (1.81) rocketed once again to 2019 levels, while he’s striking out batters at approximately the same rate as 2020 (19.8 percent), he’s walking fewer batters again to some extent (13 percent). Those two numbers have resulted in a 6.9 K/BB ratio.


To briefly spend some time on Glasnow, he never quite found his way in Pittsburgh. Although his stuff was obviously good just by appearance, his results never reflected it. He carried a 4.24 ERA in 2016 and a 7.69 mark in 2017. While other metrics fared better, like his 4.26 and 6.30 FIP, 4.57 and 5.64 xFIP over those two seasons, the Pirates — or perhaps Glasnow himself — weren’t able to tap into that next level that the pitcher seemed poised to get to.

In his first full season with Tampa Bay (2018), Glasnow pitched to a 1.78 ERA over 12 games and accumulated 2.3 fWAR. Whatever Glasnow wasn’t achieving in Pittsburgh, he was finding in South Florida. Cue the miserable Pirates’ fans and add another name to the list of squandered talent.


Since Glasnow added a slider to his repertoire in 2021, these two pitchers now employ the same four pitches as one another, and in similar numbers: 4-seam fastball, slider, curveball, changeup.

Mitch Keller’s pitch usage (Figure 1)

On the chart above (Figure 1), you can see that Keller predominantly relies on his fastball, which he throws 58.6 percent of the time. He utilizes the slider and curveball in similar measure, with the former being the secondary pitch of choice 20.1 percent of the time and the latter being preferred 17.4 percent of the time. Finally, he’s thrown his changeup 21 times this season, or, as it works out, about 3.8 percent of the time.

Tyler Glasnow’s pitch usage (Figure 2)

Comparatively, in Figure 2 you can see that Glasnow throws his fastball a little over half the time (53.1 percent), while he uses the slider 30 percent of the time, which, as you recall, is a new addition to his available options; he flips that handsome curveball in 13.9 percent of the time, and, like Keller, utilizes his changeup approximately three percent of the time.

Before moving on, there’s one fairly obvious thing to note: velocity differences. Combining all of Glasnow’s pitches, he throws for an average velocity of 90.1 miles per hour, while Keller comes in at 87.8 miles per hour (-2.3 mph), with Glasnow’s fastball exceeding Keller’s by 2.7 mph and his curveball outpacing Keller’s by a massive 5.1 mph. The other two pitches, the slider and changeup, see relatively small differences (slider = -0.2; changeup = -0.9).

Even though velocity has become a massive part of today’s game, it isn’t everything. So I wanted to move onto pitch movement.

Because Keller has seen significant regression in terms of predictive analytics and because Glasnow has seen such an uptick in both predictive metrics and real output, I first wondered if there had been a significant decrease in overall movement for the former and an increase in movement for the latter. First, let’s look at horizontal break:

(Figure 3)

Figure 3 examines Keller’s average horizontal break over the last three seasons. Keller’s pitches — excluding the curveball — have mostly remained the same. His changeup currently resides at 13.9 inches, the fastball at 6.6 inches, and the slider has risen slightly to 4.4 inches. The curveball has gone from 9.2 to 10.5 inches.

That’s all well and good, I’m sure you’re thinking, but how does Keller’s horizontal movement compare to Glasnow’s (and how does it compare to the average)? Let’s look.

(Figure 4)

Figure 4 shows a fluctuating output from Glasnow over the last several years. We’ve seen his changeup significantly alter its path positively, going from 7.4 inches of average horizontal movement to 12.7 inches, while his curveball has declined from 7.4 inches to 4.1 inches. What’s difficult to see here is Glasnow’s slider movement, which has only been in his repertoire during the current 2021 season, carries 2.6 inches of horizontal movement.

To compare these two pitchers’ horizontal movements versus the average, I present these two tables:

Figure 5: Mitch Keller’s horizontal movement vs. average (2021)
Figure 6: Tyler Glasnow’s horizontal movement vs. average (2021)

At a glance, Keller’s overall movement profile is better than Glasnow’s, at least in terms of horizontal movement. But as far as the two pitches these two pitchers throw the most, both have horizontal movement that’s below average relative to their peers. Those two pitches, the fastball and slider, are two pitches which benefit greatly, in part, from horizontal movement.

Let’s briefly discuss pitch location for fastballs and sliders.

Figure 7: Mitch Keller fastball location (2021)
Figure 8: Mitch Keller slider location (2021)

Figure 7 takes a look at Keller’s fastball location, while Figure 8 takes a look at Keller’s slider location. Keller’s fastball has averaged 94.4 mph throughout the 2021 season. It has become increasingly common in baseball for pitchers to work fastballs up in the zone, particularly with high spin rates.

Keller has averaged 2,355 rpm on his fastball this season, which results in 24.9 Bauer Units, which is approximately one unit above average. For fastballs, the higher the Bauer Units, the more the ball appears to “ride,” which results in more swings underneath the ball, making working up in the zone the correct option. The opposition is batting .230 against the fastball with a .342 wOBA.

As for the slider, Ben Clemens of FanGraphs wrote an interesting article attempting to parse the data surrounding what makes certain sliders effective.

Figure 9

Looking at Figure 9, Keller falls into the category of “87+ velo” and “2.5-4.8” inches break, which produces a run value of -1.07, which, on this particular table, is the third best outcome behind the -2.50 and -1.45 run values. Thanks to his velocity, Glasnow also falls into this category, meaning, all things considered, both pitchers’ sliders work similarly well — at least in theory.

Figure 10

Finally, with regard to sliders, it benefits all pitchers the most to work in the shadow of the plate, i.e. working the corners. This isn’t revelatory, and pitchers have been trying to do this as far back as we care to consider. Taking a look at that table, the amount of movement that both Keller and Glasnow fall into works out to a -2.60 run value for hitters in the shadow portion of the plate.

Keller’s map in Figure 8 shows that he generally works in the shadow of the zone with his slider. Going back over his last several starts, he throws approximately double the amount of sliders in the shadow part of the zone versus the heart of the zone, with the notable exception of the Tigers’ game April 22, when he threw two more “heart” sliders than “shadow” sliders.

Keller’s slider velocity has averaged 87.2 miles per hour this season, with a 2,405 spin rate, which results in 27.6 Bauer Units.

Now we can take a look at Glasnow:

Figure 11: Tyler Glasnow fastball location (2021)
Figure 12: Tyler Glasnow slider location (2021)

As can be seen in Figure 11, Glasnow works heavily in the zone with his fastball, challenging batters to try to hit it (while this is true for his time in Pittsburgh, as well, he spent much more time throwing fastballs down the middle of the plate). Glasnow’s fastball has averaged 97.1 miles per hour this season, which is the highest of his career, in addition to an average spin of 2,437 rpm. That results in 25.1 Bauer Units, which is marginally higher than Keller’s, but both pitchers should see the benefit of “ride” on their pitches.

As Glasnow’s slider chart shows, he also prefers to work in the shadow of the zone, although his ratio of shadow-to-heart isn’t as high as Keller’s over each player’s last five starts. But one thing to note is that Glasnow appears to work his slider in the middle, outer-third part of the zone, which should theoretically work well in tandem with his often-in-the-zone fastball. Lastly, Glasnow’s slider has averaged 87.7 miles per hour with a 2,788 rpm spin rate, resulting in 31.8 Bauer Units, significantly higher than Keller’s.

Of note, Glasnow produces a vertical break of 39.4 inches on his slider (5.5 inches better than average), while Keller produces a vertical break of 34.6 inches (1.3 inches better than average). This could also help explain Glasnow’s newfound success with his slider, because although it’s marginally running away from hitters, it’s dipping under their bats.

Those facts alone could go a long way in why Glasnow is producing a 43.1 percent whiff rate on the slider, while Keller’s slider only boasts a 24.7 percent whiff rate. Furthermore, although the Bauer Units are similar for Keller and Glasnow’s respective fastballs, Glasnow’s nearly three miles per hour greater velocity difference helps to result in his 29.8 percent whiff rate on that pitch versus Keller’s 17.9 percent whiff rate.

To briefly touch on each pitcher’s curveball, both Glasnow and Keller throw their curveball between 13-17 percent of the time, but see vastly different outcomes. These are perhaps their least similar pitches. Keller averages 78.6 miles per hour and a spin rate of 2,670 rpm (34 Bauer Units), while Glasnow averages 83.7 miles per hour and a spin rate of 3,033 rpm, which is 94th percentile (36.2 Bauer Units). The batting average against Keller’s curve is .333 and produces a 16.7 percent whiff rate, while the batting average against Glasnow’s curve is .056 and produces a massive 58.7 whiff rate.

Interestingly, however, is that, according to Baseball Savant, Keller’s curveball actually produces more vertical break (60.4 inches), which is up nearly four inches from 2019. Glasnow, on the other hand, produces 57.5 inches of vertical break. Both pitchers experience vertical break 5.4-5.5 inches better than average.


To conclude, while it’s true that Glasnow outpaces Keller in certain metrics, like spin rate and velocity, these two pitchers like to attack similarly, opting for fastballs in the middle-to-upper part of the zone, sliders chasing the shadow of the plate, and even utilizing each pitch in similar measure. Furthermore, each pitch that the two throw share similar Bauer Units.

But, as has been made evident, Glasnow does throw harder; he also produces greater spin and whiff rates. That fact could lead us to continue giving up on Keller, although I wouldn’t be willing to do so just yet.

If you take a look at Glasnow’s output during his Pirates’ days, some of his metrics more closely match Keller’s. For example, Glasnow wasn’t throwing as hard on average in Pittsburgh. In 2017, he averaged 94.6 miles per hour on his fastball and 79.9 miles per hour on his curveball. Both of those metrics are higher than current Keller, certainly, but not nearly as much as they are now.

In addition to his velocity spikes, Glasnow has also experienced spin spikes as well. In 2017, Glasnow’s fastball spun at 2,236 rpm, while his curveball spun at 2,571 rpm on average. Both of those marks are lower than Keller’s current output. The final difference is that Glasnow was slightly younger than Keller during that 2017 season.

Far be it from me to suggest that Keller is going to match what Glasnow has done in Tampa Bay, but I do think it’s entirely possible for Keller to step into his own as a pitcher. The Pirates aren’t short on time as an organization right now, and giving Keller the proper planning and opportunity is key to his development. He might never make it to where Pirates’ fans hoped he would, but there’s no reason to give up on him yet.