How workload data can help optimize a college pitching staff
So much is optimized in modern baseball.
Velocity-building programs have proliferated throughout amateur and pro baseball, with Driveline playing a leading role.
Pitchers have created designer breaking balls and sharpened existing offerings through advances in pitch design, another skill-building breakthrough pioneered at our Seattle facility.
Hitters have found they can increase bat speed, improve swing planes, and hone plate discipline.
These are all great breakthroughs, advancements that were once not thought to be accessible.
But one area that is far from optimized is pitcher usage and workload optimization.
In the first two installments of our workload management series, we have focused on the pro game in studying Nolan Ryan and Yoshinobu Yamamoto. But we now want to turn our attention to the college game. The best college arms are not always employed as often as they could be or deployed in the best roles for their skills. More can take on additional innings, more pitches, more leverage – helping their programs while building their prospect profiles.
For so long, coaches and players were searching in the dark for answers as it pertains to how to manage workloads and innings, and how to build up to maximum workloads. But with Driveline’s PULSE technology and the workload metrics it informs, we now have a much better understanding of how to build up workloads, and how to manage them safely.
In this third installment of the series, we will examine how to optimize the college game from the starting rotation to the end game.
Optimizing relievers
You may not be familiar with David Berg, but he authored one of the most remarkable college baseball seasons on record in 2013.
The Pac-12 Pitcher of the Year that season led the nation in saves (24), which was also good for a Division I record. The right-handed submariner-style reliever also finished second in ERA (0.92), which was first among power conference schools.
He helped the club to its first and only College World Series title in 2013, something the program failed to achieve when it had both Gerrit Cole and Trevor Bauer in its rotation just a few seasons earlier.
And perhaps most remarkable was the volume he produced.
He appeared in 51 of the Bruins’ 66 games in 2013, a tally which tied the NCAA Division I record – matching him with Connor Falkenbach (Florida, 2005).
He was not a one-year wonder of capacity, either. In 2012, Berg appeared in 50 games, which still stands as the third best mark on record.
In fact, no reliever has exceeded 47 appearances since Berg’s 2013 campaign.
For perspective, if we extrapolated his record workload over an entire, major league season, Berg would have appeared in 125 games and pitched 191 innings.
The only major league pitcher to make 90 appearances in a season is Mike Marshall, who did it twice: 92 in 1973 and 106 in 1974.
Replicating Marshall’s workload share in the pro game might be about impossible – it would certainly be pushing into uncharted territory this century. But given the number of off days during a college season, those additional periods of rest mean top relief pitchers ought to be able to appear in a greater share of games, and absorb a greater share of innings.
Josh Hejka is a Driveline researcher and a minor league pitcher in the Rangers organization. As a former pitcher at Johns Hopkins, he believes the beauty of optimizing bullpens is it would help most schools at the college level – not just the elites.
Ideally, Hejka said a high-leverage reliever ought to be utilized more frequently and not pigeonholed into a set role, such as only pitching the ninth inning in a safe situation. This would allow for more appearances, and a greater share of high-leverage performance.
While Berg appeared in 77% of games during UCLA’s 2013 season, typically a high-end Division I reliever appears in 35-50% of games, according to a ChatGPT analysis of Division I stats.
“I do think the best reliever probably should be utilized more in a way like the Andrew Miller prototype rather than the closer prototype,” Hejka said.
It should be possible for a top fireman, pitching in shorter-duration outings, to appear at least three times a week in games. That would push appearances up to a Berg-like range.
But there’s a need for a second relief ace archetype, one who can take on a greater volume of innings per appearance.
Hejka, also a submarine-style arm, also pitched a high volume of innings for Johns Hopkins. He pitched in a regular, high-volume role. He once tossed 150 pitches in a Division III College World Series game in 2019, and recorded a save two games after that appearance.
“I would come in during the fifth inning and finish the game, if it was a close game,” Hejka said. “I think there’s room to use your best relievers in that way because of the built-in rest time. You can throw them for more innings.
“Maybe there’s room for a shorter-appearance guy that can pitch every single day, every Friday, Saturday, and even Sunday if he needs to. Two innings max in an outing. And then you have a guy that can come into a game on a Saturday and pitch the fifth, sixth, seventh, eighth and ninth, and then be good again on Tuesday to go three or four more. That’s really, really powerful. It’s basically guaranteeing you a win in a close game.
“You’re maximizing the value of that arm.”
Consider that most Division I teams allocate about 40-50% of innings to their weekend starting pitchers. The secondary starters pitching midweek games and/or covering for injury will account for another 15-20% of innings. Let’s call it 60% of innings allotted to starting pitchers.
That leaves 40% of innings to be picked up by relievers, but too often those innings are not allocated in an optimal manner.
Driveline’s Max Engelbrekt is a former Oregon State pitcher, who like Hejka believes many programs are far from employing optimized staffs.
Instead of so much fragmentation within the remaining 40% of innings, Engelbrekt said if a team could concentrate half of that into its best relief arms, it would enjoy a great advantage. While there’s often talk about a shortage of arms in the pro game, college staffs are stretched to a much greater, relative degree.
“The level of depth on MLB teams is much narrower in talent density than with top-end starters in college,” Engelbrekt said. “So it just becomes that much more important to make sure you’re throwing your leverage arms as much as they effectively can.”
Building a pair of ace relievers requires greater but different workloads. But it is possible with proper planning and training to build such arms. And with PULSE and workload units we can do just that – college baseball doesn’t have to be searching in the dark.
Let us begin with a key, PULSE-informed metric that guides us: acute-chronic workload ratio (ACWR). This workload guardrail compares an athlete’s training load over a brief period (acute), a week for a pitcher, to their workload over a longer period (chronic): the previous 28 days for a pitcher in our methodology.
The concept of ACWR began outside of baseball, and research extends to a number of sports, too. Most athletic disciplines share a common finding: an athlete should increase their acute workload to build capacity, but not too quickly. We want to keep an ACWR value at 1.2 or lower, an acute value that does not exceed the chronic by more than 20%.
PULSE and ACWR can help guide us to safer and smarter workload management and buildup.
Consider Hejka’s simulation for a one-inning, regular fireman:
And for a higher-volume reliever:
“Sometimes you’ll see these top relievers up to like 60 to 70 innings and that’s all in relief, which is pretty damn cool,” Engelbrekt said. “That way you can almost eke out like starter innings.”
We can build more Bergs – and Hejkas and Engelbrekts. We can optimize college bullpens.
Optimizing starters
When we see some of the elevated pitch counts that starting pitchers tally, often seen in the NCAA Tournament, it can open eyes and create headlines.
After all, aren’t we supposed to be focused on pitcher arm health, especially amateur pitchers?
Way back when I covered Clemson University in the early 2010s, the University of North Carolina sports information official asked the press not to report Tar Heels ace Matt Harvey’s 157 pitches on the day, an alarming number to those of us in the Doug Kingsmore Stadium press box. We reported on it. It seemed reckless to me at the time, but perhaps he was equipped for it.
His 157th pitch still traveled 96 mph, and he went on to have a brief but dazzling peak in the majors.
With more data to study, with improved monitoring, it is possible that such single-start workloads can be safe and effective in the college game. Like with bullpen arms, the best starting pitchers can also take on more innings.
But to even consider this, pitchers have to build the capacity to handle more work in a start. Earlier this year, Hejka modeled what building a Nolan Ryan-like workload would entail.
“That is probably pretty similar to the Nolan Ryan (simulation), but just with more space in between,” Hejka said. “It’s probably a little bit more doable at the college level.”
Here’s what applying the Ryan-like model to a college starter would look like:
In the last five college seasons, only four starters have reached 120 innings in a season topped by Quinn Mathews’ 124 innings with Stanford in 2023. Paul Skenes is second with his 122 2/3 innings with LSU, also in 2023.
More innings have been thrown at the college level in previous eras.
For instance, Floyd Bannister tossed a Division I record 186 innings for Arizona State. We are not advocating that to be a standard, but there is potential for much more capacity. Perhaps some of us should not be so alarmed by elevated, single-game pitch counts – assuming the pitchers are building up, and being monitored, responsibly.
There is another model to consider, too: the starter who can pitch more than once per week. The typical college starter starts once per week.
Consider Scott Erickson of the 1989 Arizona Wildcats. Erickson started 20 of the club’s 64 games, firing 14 complete games, and appearing in another four games in relief.
He managed to start 31% of the team’s games, tossing a school record 172 innings, which ranks eighth all-time.
How could teams get more usage out of their top starting pitchers?
It’s difficult to copy a four-days-of-rest schedule in college baseball given the typical schedule format of three weekend games (Friday, Saturday, Sunday) and one midweek game (Tuesday or Wednesday) during the season.
But a creative club could squeeze in more work for its best arms.
If LSU wanted to get more from Skenes in 2023, and he was restricted to never pitching on fewer than four days’ rest, he could have picked up two additional starts, moving from 19 to 21 on the season. Extrapolating his per-start average, that would have yielded an additional 13 innings.
Now, for a power program like LSU, it might not be necessary to try and increase start volume from top starting pitchers during the regular season.
But other programs, like mid-majors, extracting those extra innings, that extra value, could be the difference between making the tournament or not.
And that’s most programs.
“I would argue it’s the lower-level schools that need to get extra work from their existing best,” Hejka said. “If you’re LSU, you probably don’t need to fully optimize every reliever, and every starter to make the tournament … But those mid-majors that are trying to push for a tournament berth, those are the guys that maybe could benefit from (optimizing a staff). Start your best arm on Friday, bringing back Tuesday or whatever.
“A smaller school might get a couple extra wins out of the season by being able to optimize stuff like that. So, I think it’s aiming towards the majority rather than the minority of teams. Not only are those the teams that need it the most, it’s also that more teams, more people, would benefit.”
Every team could benefit from getting more out of their best arms and using the college schedule and its additional off days. We’ve optimized much in modern baseball, but this is an opportunity that enterprising college programs can exploit.
Comment section