Since taking over as Twins manager prior to the 2019 season, Rocco Baldelli has frustrated many fans with how early he removes his starting pitchers from games. Let’s travel through time and observe three starts from the past couple years. 

June 18th, 2022 - Twins at Diamondbacks (Box Score)

Starting Pitcher: Dylan Bundy

This is Bundy's 11th start of the season since signing a free agent contract with the Twins for one year and $4 million. Up to this point, he'd only been allowed to start the 6th inning four times out of 10 starts and had maxed out at 6 innings pitched. With a 5.87 ERA (thanks to 13 earned runs in 12 1/3 innings pitched over his previous three starts), he was looking to turn his season around in Arizona.

Well, going 8 innings with only 1 run allowed in an 11-1 victory appeared to be a good sign. The D-Backs couldn't muster much against his 107 pitches, getting only 4 baserunners while striking out 7 times. Surely this meant Bundy had earned some more trust from manager Rocco Baldelli, but that proved to be false as Bundy was only allowed to throw 60 pitches and 6 innings at home against the Rockies in his next start (a 1-0 loss). After that loss to the Rockies, he never completed 6 innings again and only topped 80 pitches twice for the remainder of the season, his final year in MLB. 

June 8th, 2024 - Twins at Pirates (Box Score)

Starting Pitcher: Simeon Woods Richardson

Woods Richardson started the season in the minor leagues and started one game as part of an April 13th doubleheader before joining the rotation full time thanks to Louie Varland's demotion. After a cup of coffee in each of the prior two seasons, this was SWR's first extended look in the majors and he seized what Varland could not. Even with a 5-run hiccup against the Blue Jays nearly a month earlier, Woods Richardson sported a tidy 3.05 ERA entering this game and the Twins were starting to reap the rewards from trading away Jose Berrios several years earlier. However, Woods Richardson was being treated much like Bundy two years earlier, as he wasn't being pushed beyond 6 innings and was often removed early despite allowing 2 runs or less.

This start in Pittsburgh was a different story, though, as Woods Richardson stood atop the pitching mound to kick off the bottom of the 7th inning. Nine times prior, SWR was off the field at this point, but Baldelli entrusted him to maintain a 0-0 tie through 83 pitches. Unfortunately, Baldelli's faith did not pay off, as Woods Richardson allowed a home run to Rowdy Tellez, and the Twins were unable to score in what was ultimately a 4-0 loss. 

June 9th, 2023 - Twins at Blue Jays (Box Score)

Starting Pitcher: Sonny Gray

I swear it's purely coincidental that these are all away games in June, but if anyone would represent Rocco Baldelli's propensity for quick hooks for starting pitchers, it had to be Sonny Douglas Gray. We're familiar with Gray's frustration of being pulled earlier than he preferred while with the Twins, and this day featured exactly that. He entered this game with a 2.28 ERA on the season and had shut down the Blue Jays through the first 4 innings. His 5th wasn't as good, needing 31 pitches while allowing 4 baserunners and his first and only run of the game, but 76 pitches and a 2-1 lead at the end of the 5th was still laudable. However, Gray was forced to watch Jovani Moran give up the tying run in the 6th inning, and the outcome would have been more sour if not for the Twins plating the eventual winning run in the top of the 10th inning. 

---

Okay, let's return to the present. Just like you, I've noticed the demise of the starting pitcher in Twins Territory. Complete games? Gone. Cresting the 100-pitch threshold? Gone. Your sanity as Rocco Baldelli makes his somber march to the pitcher's mound in the 5th inning? Gone, gone, gone. Even with his pitching staff finishing 4th in MLB in innings per start in 2023 (5.52) and 14th in 2024 (5.26), he receives frequent criticism for pulling his starters too early in games. (Being 21st at 5.14 innings per start through the All-Star break this season isn’t helping matters.) 

We're aware of the struggles pitchers endure while facing the opposing lineup a third time in a game, and pitchers have conga-lined their way to Driveline, Tread, etc. to throw harder and learn nastier pitches to the detriment of their in-game stamina. This has caused fans to be so frustrated that we now have proposals for the double-hook (the starting pitcher leaving with less than 5 innings pitched causes his team to lose their DH for the rest of the game) and stricter roster limits, like a maximum of 11 pitchers on the active roster. 

However, sometimes Baldelli breaks from tradition and allows some starting pitchers to pitch deeper into games. Alternately, he also removes his starters early when they appear to be cruising, such as Woods Richardson against the Cubs Tuesday last week (0 runs allowed on 2 hits through 5 innings, but only 61 pitches thrown). It hasn't felt that there's been a discernible reason when some pitchers get to break free from the organization's philosophical grasp, so I wanted to identify when Baldelli and Co. choose to remove their starting pitcher. 

You see, the Bundy and Woods Richardson starts mentioned at the top stood out to me as a Twins fan. I recall fellow fans balking at Bundy getting pulled after 5 innings and 1 run allowed, but the counterargument was, "It's Dylan (expletive) Bundy! You really want to see him pitch more?" Essentially, you had to assume he was going to be mediocre each start. Anything better than that, Baldelli should cash in his chips before the odds started swinging the other way. Yet that start in Arizona piqued my interest because, sure, it was 9-0 after 3 1/2 innings, but Bundy was allowed to pitch far longer than any other start that season. It's not like his performance was unusual, as he had 10 other starts that season where he allowed 1 run or less, but the Arizona start was 2 full innings longer than his next longest start, which occurred twice (one of which was his following start against Colorado). 

Likewise, Woods Richardson was often pulled despite pitching more than adequately. It seemed automatic, he'd be allowed to complete 5-6 innings, but he was typically removed before the 7th inning despite being nowhere near 100 pitches. From watching Twins games daily, I noticed that Baldelli would often turn to the bullpen early if the score was close. That start in Pittsburgh was different, though, as SWR was handed the 7th inning in a scoreless ballgame. It ultimately backfired as Woods Richardson gave up the game's first run while facing the Pirates' lineup a third time, but it was interesting that this was the time that he finally got to pitch into the 7th inning.

Therefore, I started my research shortly after that Woods Richardson start in Pittsburgh. I began by identifying the Twins pitchers in the Baldelli era that seemed to get frequent quick hooks despite pitching well. Bundy, Randy Dobnak, and Devin Smeltzer jumped out to me immediately. The Pittsburgh start for Woods Richardson was the whole reason I started this project, so he was in. Matt Shoemaker got the Bundy treatment, too, he was in. Same with Kenta Maeda, Rich Hill, and Sonny Gray.  Louie Varland? Come on down! Fans were mad about Dallas Keuchel's first Twins start (5 innings, 8 hits, 0 strikeouts, but only 1 run allowed)? Save him a seat. 

But now I had a problem. I accumulated all these pitchers that potentially had flaws that prevented them from pitching deeper into games, but there wasn't a control group, so to speak. Thus, I added in the top three pitchers in recent seasons, Pablo López, Joe Ryan, and Bailey Ober. I also found it interesting that J.A. Happ seemed to pitch into the 7th inning more frequently than most of the other pitchers, so he was added in as well. 

This gave me a total of 14 pitchers and 425 starts ranging from May 28th, 2019 to June 16th, 2024. I want to note that this does not include every pitcher that made a start for the Twins, as I focused only on those that seemed to be on short leashes unfairly and those that were the absolute best. For example, I don't think any of you are upset that Josh Winder averaged under 5 innings a start in 2022, or why Charlie Barnes' 8 starts in 2021 failed to make the cut. I also omitted Chris Archer in 2022, as complications from his surgery for thoracic outlet syndrome prevented him from pitching more than 4-5 innings per start. The same is true for Bailey Ober's 2021, as the Twins were open about keeping his innings near his 2019 career high of 78 2/3 after the 2020 minor league season was cancelled due to the COVID-19 pandemic. Any other outings that were cut short due to injuries, illness, rain, poor performance, etc. or were “bulk” outings following an opener were also removed as they didn’t fit the spirit of the exercise. This pared down the number of starts to 335. 

Next, I had to determine what stats to use. I started with innings pitched, batters faced, and pitch count as measurements of start length. As for measuring performance, the standard box score stats seemed obvious (hits, runs allowed, earned runs allowed, walks, and strikeouts) but also rudimentary. One run allowed in Bundy's 11-1 win versus Arizona was much different than Gray leaving his White Sox start ahead 2-1, so I also looked at the score when the pitcher was removed and the game's final score, along with which opposing batters were due up later in the inning or in the next inning. Matthew Trueblood gave me a good suggestion as well of looking at the game’s leverage index (how stressful a situation is based on the score, inning, number of outs, and runners on base, where 0 is least stressful, 1 is average stress, and anything above is higher stress) when the pitcher was removed to determine if the current game situation warranted the pitcher’s removal. 

Most of this data still felt very basic, but I did have a remedy. Though I don't recall the exact quote, I remember the Twins broadcast noted during Woods Richardson’s remarkable start against the Pirates that he generated many swings and misses. Indeed, he ended the game with a 16.5% whiff rate, his second-highest in a start up to that point. (Note: It appears that FanGraphs and Baseball Reference have different numbers for whiff rates, so FanGraphs says it’s 15.5% while Baseball Reference has 16.5%. I believe the difference is regarding whether to include foul tips as swinging strikes.) So, I started keeping track of swinging strike percentage, along with a pitcher's called strike plus whiff percentage (CSW%). Additionally, I wanted a statistic for overall offensive performance and expected offensive performance, so I recorded each start's weighted on-base average (wOBA) and expected weighted on-base average (xwOBA). I figured that while a pitcher's surface-level stats like earned runs allowed might look good, perhaps he had allowed a lot of hard contact and the Twins wanted him out of a game before his luck turned. 

Lastly, since we hear so often about how a pitcher fares against a lineup the third time through the order, I recorded each player’s tOPS+ allowed when facing a lineup a third time, or their OPS allowed the third time through compared to their overall OPS allowed. A tOPS+ of 95 when facing the batting order a third time means that pitcher performed (100 - 95)% or 5% better than his overall OPS, while a tOPS+ of 105 means he performed (105 - 100)% or 5% worse than his overall production. 

Using all of this information, I then reviewed each start over the time period for these 14 pitchers and determined whether I agreed, partially agreed, or disagreed with when Baldelli removed the pitcher from his start. I will admit, I didn't pay any attention to who was available in the bullpen for each start, as reliever availability is due to a myriad of reasons beyond just "how many pitches has he thrown recently?" and I felt it would be too difficult to determine who was truly available each day by reviewing the previous days’ games. I also did not create an objective measure for differentiating a good hook versus a bad hook, resorting instead to what the kids these days call "vibes." 

After all that work, I identified my results. If you'd like to peruse the data yourself, you can find it here. Green rows mean I agreed with the hook, yellow means I partially agreed, red means I disagreed, and purple is an omitted start due to injury, illness, bad performance, etc. Some rows have blue text, which means I felt they were the “best” hooks made by Baldelli. Others are boxed, which were notable games that could have been featured at the top of this article. I also included some notes for nearly every game if I felt it was relevant. 

Here are significant findings from this study. 

1) It seems that many fans still believe 100 pitches is the threshold for starting pitchers to reach and that hasn’t been the case for a long time. Many of the starts ended with the pitcher throwing about 80-90 pitches, likely because starting a new inning would have guaranteed the pitcher would get to 100 or more. A quick inning would have been fine, sure, but there was a risk the starter would struggle, leading to a reliever coming into a tight spot mid-inning. I know that teams would prefer to give their reliever a clean inning, thus they err on taking the starter out a little early. Plus, the lower pitch counts are a current strategy in the hopes of reducing pitcher injuries. (Whether or not it’s working is debatable.)

2) You've likely noticed that the Twins rotations have been righthander-heavy for years now. If a pitcher was reaching the 80-90 pitch threshold with lefthanded batters coming up and Caleb Thielbar available, Thielbar was coming in. This season, we’re seeing a similar operation with Danny Coulombe. Going back to last Tuesday’s game against the Cubs, Woods Richardson was pulled after 5 innings in favor of Coulombe with three of the next four batters hitting lefthanded. The idea is like Baldelli’s pinch-hitting strategy - use your bench/bullpen for the present situation, rather than hoping for a similar situation later in the game. If he didn’t use Coulombe in the 6th against the trio of Cubs lefties, the next time those batters would hit would be the 8th or 9th inning - situations typically reserved for Griffin Jax and/or Jhoan Duran.  

3) I found the correlation coefficient for the three measures of start length (innings pitched, pitches thrown, and batters faced) compared to various performance stats and leverage index at time of removal. The correlation coefficient r ranges from -1 to 1, where 0 is no correlation between the variables, -1 means a perfect negative correlation (as one variable goes up, the other goes down) and 1 is a perfect positive correlation (as one variable goes up, the other also goes up). 

As you can see, only a couple pairs had as much as a moderate correlation: innings pitched vs. leverage index when removed, IP vs. wOBA, IP vs. xwOBA, and batters faced vs. leverage index when removed. Every other pair had either a weak or virtually no correlation.

The correlation between wOBA and innings pitched makes sense, you’re going to be allowed to pitch deeper if you’re shutting down the opposing offense. Likewise with leverage index when removed and batters faced or innings pitched, as you’ll get to face more batters and go deeper into the game if the score is lopsided (in the Twins’ favor, as the previous correlation shows). In fact, Baldelli even confirmed that immediately after the Tuesday Cubs game, where Woods Richardson was removed after only 61 pitches. Of course, these are only moderate correlations, though, so these only partially explain how long a Twins starting pitcher will go in any given start. 

—

After all of this research, I’m left both satisfied and unfulfilled. While hoping to find the single determining factor was an ambitious task as to when Twins starting pitchers are removed from games, I still feel pleased that I found a couple answers. Score and leverage index seem to be the primary factors, along with the availability of top lefthanded relievers. However, I know that I could have used or created an objective measure of start quality that likely would have made my findings more accurate, and I’m sure I could have dug up more pitching stats that perhaps would have yielded better correlations with start length. 

Regardless, my main goal was to give Twins Territory a better understanding of when Rocco Baldelli chooses to remove his starting pitchers. I can’t promise that this will completely lower your blood pressure, but hopefully this gives you some more answers to his thought process.