Which Randy Dobnak should we expect in 2021?

One of the Twins’ position groups that has been most in need of roster additions this offseason is the starting rotation. Kenta Maeda, José Berríos, and Michael Pineda are locked into the first three spots and the free agent signing of J.A. Happ seems to have filled the fourth spot. But with Jake Odorizzi, Rich Hill, and Homer Bailey allowed to become free agents, the fifth spot is uncertain. Many Twins fans and writers are holding out hope that Falvey and Levine make another free agent investment in the starting rotation instead of filling it internally.

Adding a free agent starter would fortify the Twins’ rotation depth and push Randy Dobnak, Devin Smeltzer, and Lewis Thorpe to depth spot starter / swing man roles. Of those three, Dobnak would appear to have the inside track to a rotation spot if a free agent is not added.

The sentiment that the Twins would be better off with a free agent fifth starter instead of Dobnak is a reflection of the Twins’ place inside a competitive window, outside pressure to keep pace with a rival (Chicago) that has invested heavily in its roster this winter, and the Front Office’s clear commitment to building and maintaining roster depth.

It also reflects our collective uncertainty about Randy Dobnak.

Dobnak’s final 2020 line was 6-4, 4.05 ERA, 3.96 FIP, and 0.8 fWAR in 46.2 innings. Altogether, he was about 10% better than league average (91 ERA- and 89 FIP-) in 2020 and through 19 career appearances (15 starts), he has delivered 3.12 ERA and 3.56 FIP. That grades out somewhere around 20% to 30% better than league average (69 ERA-, 79 FIP-) and is production that every team would gleefully take from its fifth starter.

Yet, we mostly don’t know what to make of Dobnak. His unconventional path to the Major Leagues and lack of prospect pedigree casts a shadow on his legitimacy and staying power as a big league starter. His dependence on inducing ground ball contact instead of missing bats also creates pause. The fact that his excellent numbers above have come in just 75 innings across two small samples at the end of 2019 and a shortened 2020, makes them hard to trust. And all of those factors are exacerbated by the fact that Dobnak’s 2020 season, despite his solid overall numbers, ended with him being optioned to the alternate training site after a string of rough starts.

So here we are, a few weeks before Spring Training is supposed to begin, and we have what appears to be (by the numbers) a luxury of a fifth starter that a lot of people would prefer wasn’t the team’s first option as the fifth starter. Given all that, it seems like a worthwhile task to try to figure him out. As Twinkie Town regular geomyidae asked recently, “is he for real or not?”

2020 Before and After

The 2020 season totals I showed above mask a tale of two seasons for Dobnak. First were five fantastic starts to begin the season that had analysts like me calling him a “mid-rotation starter” and then five mostly miserable starts that resulted in him getting optioned out.

Setting aside the obligatory caveat about being cautious in discerning insights from the odd 2020 season, it makes sense to start with a before and after look at Dobnak’s season.

Through five starts (25.1 IP, 96 batters faced), Dobnak allowed just four runs (1.42 ERA) and held opposing batters to a .178 / .229 / .278 triple slash line that yielded only .225 wOBA. In the five starts that followed (21.1 IP, 104 batters faced) he allowed 17 runs (7.17 ERA) and allowed opposing batters to hit .378 / .441 / .456, which worked out to .395 wOBA.

Let’s dig into the peripheral stats to see if anything stands out as a change between the two stretches:

The stats believed to be within Dobnak’s control — walks (BB%), strikeouts (K%), and home runs (HR/9) — mostly stayed the same. Remember that we’re working with small samples here, so the slight differences in those three categories in the table are just noise — only 1 more walk and home run allowed and just 1 fewer strikeout. As a result, the fielding independent pitching (FIP) marks between the two samples are mostly similar.

Where there were big changes were in batting average allowed on balls in play (BABIP) and runners left of base percentage (LOB%) — both of which were demonstrably worse in his last five starts. It’s worth pointing out that last season’s league averages for BABIP and left on base percentage were .291 and 71.8%, respectively. Dobnak clearly experienced both sides of the fluky luck coin in his 10 starts.

Next, let’s look into the comparison of Dobnak’s batted ball profile:

Here it’s more of a mixed bag. In both stretches he generated a very high percentage of ground balls and mostly kept the ball out of the air. In the last five starts, a few more of those batted balls turned into line drives, but not in any kind of major way.

Altogether, Dobnak’s 62.1% ground ball rate for the season was the highest in baseball among the 111 pitchers who threw at least 40 innings last season.

Interestingly, given the disparity of results he allowed in the two stretches, the average exit velocity allowed remained steady at 87.5 mph and the average launch angle and hard hit rate improved in the second half of the season.

To put all this data together, at least in terms of things Dobnak had influence over, it looks like he was mostly the same guy during the two stretches. He got similar rates of strikeouts and ground balls, and continued to limit walks, home runs, and hard contact. Those are all good things.

From the data we’ve investigated so far, it looks like this could be a story about bad luck and small samples. But is it really that simple?

2019 vs. 2020

Here, it’s probably worth broadening the view beyond 2020 to gain some context. In comparing Dobnak’s data between 2020 and 2019 seasons there were two key differences that jumped out to me.

His strikeout and walk rates both trended in the wrong direction in 2020. In his 2019 debut campaign he struck out 19.5% and walked 4.2% of the batters he faced. The numbers in the two tables above show that he became more contact oriented last season, declining from his already below average levels of missing bats.

The second change occurred in the results he allowed against left-handed hitters. In 2019, he held the 59 lefties he faced to a .175 / .203 / .211 line that rolled up to just .183 wOBA. Those figures were aided by a very good (and unsustainable) .208 BABIP allowed versus LHB. Before you chalk it all up to good luck, know that the Statcast-derived expected metrics against lefties were pretty good, too — .266 xBA and .349 xSLG.

His success against lefties continued in the beginning of the 2020 season. In his first five starts, 51 lefties hit just .170 / .235 / .277 (.229 wOBA). Again, Dobnak was the beneficiary of some good batted ball luck, allowing just .179 BABIP. But the expected stats were still pretty decent again — .270 xBA and .403 xSLG.

After that, though, it was a very different story. The rest of the way in 2020, 40 lefties produced .471 / .525 / .618 (.487 wOBA). While the BABIP in that stretch was a ridiculous .484, much of the damage done was deserved as indicated by .335 xBA and .520 xSLG.

So maybe it wasn’t just bad luck? What might have changed that would account for the decline in missed bats and sudden inability to handle lefties?

Pitch Mix, Velocity, and Movement

An element that Dobnak has lots of control over is his pitch selection. Was there anything significantly different in his pitch mix that would give us a clue about what changed?

Not really. The chart above has the kind of variation you would expect to see game to game based on matchups and scouting, but nothing significant in terms of overall changes. Dobnak mostly operated with his sinker, slider, and changeup, mixing in the occasional four-seam fastball throughout the season. Most of the variation in the data occurs between his slider (yellow) and changeup (green) which is likely due to the handedness of the batters he faced in a given lineup — more sliders for righties and more changeups for lefties.

How about any changes in pitch velocity? Dobnak isn’t a hard thrower to begin with (31st percentile fastball velocity), so even slight changes in velocity could be a precursor to trouble.

Again, not really. If anything, Dobnak’s velocity on all his pitches ticked upward as the season went on, but by less than a single mile per hour. It doesn’t seem significant in terms of the amount of change and the increase shows up across all four of his offerings, meaning the velocity differential between his fastballs and off speed pitches remained similar.

But, maybe more velocity for a pitcher that relies heavily on downward movement isn’t a good thing? Let’s check the movement profiles on Dobnak’s pitches for any changes.

This chart is plotting the vertical movement on Dobnak’s pitches. Again, there just doesn’t seem to be anything significant here. Maybe you could squint hard and say his sinker and slider lost a little downward movement in the middle of the season, but then both clearly recovered in his final two starts.

To summarize this section, I don’t see much of anything in the pitch selection, velocity, and movement data that would explain the performance change. It seems he was still the same guy during both stretches inside the 2020 season and all of this data tracks pretty similarly to 2019, too.

If his stuff was basically the same, perhaps there were changes in his command that had an impact?

Plate Discipline, Location, and Count Leverage

Using the same kind of before and after approach as the first section above, we can dig more deeply into pitch by pitch data to investigate.

Let’s start with the plate discipline numbers:

First, let me identify the labels. O-Swing% is the percentage of pitches out of the strike zone that the batter swung at. Z-Swing% is the same but for pitches inside the strike zone. F-Strike% is the percentage of first pitches of a plate appearance that were strikes. SwStr% is the overall percentage of pitches that were swung at and missed.

A couple of things stand out. First, batters swung at fewer of Dobnak’s pitches overall during his second half struggles. The O-Swing%, Z-Swing%, and Swing% figures were all noticeably lower. That could be due to the fact that he threw fewer pitches in the strike zone, shown by the Zone% falling from 41.7% in the first five starts to just 33.9% in the last five starts. He also threw fewer first pitch strikes. All of that likely contributed to the decrease in swinging strikes on the right hand side of the table.

Perhaps we’ve found a potential culprit behind his struggles?

Let’s check the Statcast pitch by pitch location plots to confirm. Below is the plot of the 391 pitches Dobnak threw in his first five starts:

Not surprisingly for a right-handed sinkerballer, the bottom arm-side corner of the zone (down and in for the right-handed batter) is where the majority of his pitches landed early in the season. The size of the red bins inside the strike zone make it clear that Dobnak pounded the bottom of the zone.

Next is the plot of the 357 pitches thrown in his final five starts:

This one has a different pattern and the red bins are not clustered as tightly together or as clearly inside the strike zone. This makes plain that more pitches landed on the edge or outside of the strike zone during his second half struggles.

Another data point that suggests Dobnak struggled with command in his last five starts is that he hit four batters with pitches after not hitting any in his first five starts. Batters hit by pitches are often hidden by our focus on strikeouts and walks, but they are free baserunners nonetheless. If those four HBPs had been included as walks in the very first table I showed you above, his BB% in the second half would have been 10.6% — significantly different than the 6.3% from start of the season and a clear signal that his command was lagging.

Command matters because pitchers and hitters get dramatically different results when they are ahead or behind in the count. Hitting is hard as it is, but doing so when the pitcher has the count advantage is next to impossible.

For a pitcher like Dobnak, who is prone to allowing contact regardless of the count, getting the count advantage over the hitter is imperative. The numbers bear this out clearly:

The table makes it obvious. Ahead in the count Randy Dobnak is dominant. Behind in the count Randy Dobnak gets smoked.

The inconsistent command is also likely a big driver of some of the strikeouts evaporating and the performance change against left handed batters. Dobnak’s approach for lefties primarily relies on working away from them with his sinker and changeup to take advantage of those two offerings tunneling well together. The second pitch location plot above showed he struggled to consistently hit that low outside corner from lefties in the second half of the season. As a result, the sinker and changeup were likely more easily distinguished from one another by the batter and hitters could wait for Dobnak to fall behind in the count in order to get a pitch out over the plate they could hit hard.

Putting it all together

To recap, Dobnak’s stuff and approach held steady. He continued to generate bushels of ground balls and mostly suppressed hard contact while experiencing dramatic swings in batted ball luck. Those facts point to the idea that he was mostly the same pitcher we saw in 2019 and the beginning of 2020.

That idea is supported by Statcast-derived expected statistics. Dobnak’s xBA, xSLG, xwOBA, and xERA for 2020 were amazingly consistent with his 2019 numbers.

He also struggled a bit with his command as 2020 wore on, resulting in more favorable counts for the opposing hitters, a few more free baserunners, fewer strikeouts, and more hard contact from left handed batters. Those facts point to the idea that he needs to make an adjustment or two to be similarly successful going forward.

So, what’s the verdict? Is Dobnak a viable starter in 2021?

In my estimation, yes, especially as a fifth starter or depth spot starter. He likely doesn’t have the mid-rotation upside we had hoped for without some adjustments to find more strikeouts and handle left handed batters more effectively. But he is a good option for the Twins’ rotation and would be a luxury as a depth piece should the front office add another free agent starter.

The good news is the clearest adjustment he needs to make — getting back to working ahead — is achievable and in his control. In addition, Dobnak figures to be a major beneficiary of the Twins’ offseason investments in the infield defense. Perhaps more than any other Twins pitcher, Dobnak is dependent on his defense. The past two seasons, he’s pitched in front of an infield that ranked 21st and 22nd in converting ground balls into outs. With the addition of Andrelton Simmons at shortstop and resulting position changes for Jorge Polanco and Luis Arraez, the Twins infield defense figures to be much improved in 2021. That should make Randy Dobnak better too.

---

John is a contributor to Twinkie Town with an emphasis on analytics. He is a lifelong Twins fan and former college pitcher. You can follow him on Twitter @JohnFoley_21.