How Hard Are College Pitchers Worked?

Clinging to a lead in the ninth inning of College World Series Game Three, Coastal Carolina’s Alex Cunningham faced Arizona’s Ryan Haug. There were two outs and runners on second and third, so a hit would have given Arizona the walk-off 5-4 victory and its second national championship in the past five years. Instead, Cunningham struck Haug out to give Coastal its first College World Series title. The whiff officially ended the NCAA Division I season, allowing all college pitchers to finally recover.

This game featured a bevy of heavily used pitchers. Coastal starter Andrew Beckwith had thrown 138 pitches in his previous start. Bobby Holmes, the first man out of Coastal’s bullpen, had thrown 47 pitches two days prior and 39 pitches five days before. On the Arizona side, starter Bobby Dalbec had worked deep playoff outings, all after hastily leaving bullpen work midway through the year. And reliever Cameron Ming worked constantly in tournament play; he made back-to-back appearances in the final two championship games to follow a 79-pitch outing just four days earlier.

If a major league manager ran his young pitchers this hard, he’d get hell from the fans and pundits who don’t want to see star prospects abused. Given that pitching through fatigue is a chief injury risk, it would be warranted.

Some might be inclined to give Arizona coach Jay Johnson and Coastal Carolina coach Gary Gilmore a pass because this was tournament play. Maybe the seduction of a championship justifies an “all hands on deck” mentality. But these usage levels weren’t limited to the College World Series. Over the course of the NCAA regular season, regionals and super regionals this year, a bunch of ugly stories of heavy usage popped up.

Still, those individual incidents don’t quantify the extent to which overuse is a problem in college baseball. A broader perspective is needed, with all of NCAA’s D1 programs and their pitchers considered. How hard are colleges riding these developing arms?

College Baseball Data

NCAA data come with red flags. To start, there’s the issue of access—the data can’t be easily parsed through Chadwick or scraped with simple pitchRx code. Fortunately, I can rely upon the terrific scraping efforts of Beyond the Boxscore’s Bryan Cole. His Github page contains CSV files of pitching game logs, team rosters and team schedules for the 2012–2015 D1 college seasons.

Even with the hurdle of access overcome, there’s a separate issue of data reporting. Pitch counts are a preferred basis for analysis of pitcher use and abuse, but the NCAA didn’t report pitch counts for 33 percent of D1 outings. However, every game log contained each pitcher’s total batters faced. So to interpolate the missing pitch counts, I employed the following process: I found each pitchers’ yearly average of pitches per batter faced for the games in which pitch counts were reported. I took those ratios and multiplied them times the batters faced for the games in which pitch counts were missing.

Per-Start Volume

First things first: How many pitches do NCAA D1 starters throw in their outings? We’ll examine each of the four class levels, and throw in PITCHf/x data so major league pitchers can be considered as well. Appearances will be separated by season type (either regular season or tournament/postseason). Let’s use probability density functions—in which the area under each curve sums to one—to see how the five groups differ.

Comparing the panels above and below, we can see that all four groups of collegiate pitchers work harder in the tournament. The typical freshman pitch count increases from ~85 in the regular season to ~90 in tournament play. Sophomores, juniors and seniors work harder, as they usually throw ~95 pitches in the regular season and ~105 in the playoffs. Notice also that the collegians’ distributions “lean” to the right during the tournament, showing that bigger pitch counts become more common.

Major league starters are used differently. Their distributions peak markedly higher, showing how consistently they sit at 100 pitches in both the regular season and postseason. In contrast, NCAA pitch counts are much more widely distributed across the charts. Collegians do throw more short starts of 70 pitches and fewer, but they also throw a lot more starts that roll into troublesome usage territory.

A 115-pitch start in the majors is rare; they pop up in 4.4 percent of starts in the regular season, and are just a bit more frequent in the postseason. Collegians’ portion of lengthy outings is greater in the regular season and substantially greater in the playoffs. If you’re a junior, you have to figure that one-sixth of your tournament outings will span 115+ pitches. These numbers include many starts that stretch deep into the hard-to-see rightmost tails of the distributions; we’re talking about outings that can span 140, 150, 160 and even 170 (!) pitches.

In the distribution charts as well as the table, the NCAA workload gets heavier with age. Why might this be? A few possible reasons come to mind:

• Juniors and seniors are worked hard because they’re going to leave the program soon—whether it comes via the draft or graduation—and their long-term health isn’t a priority.
• Coaches trust their older, developed pitchers to do a better job, and so they allocate more innings to them.
• Coaches are protecting teenage players, who are less polished and less physically mature.

These pitch totals are troubling, so we need to dig deeper, beyond this single dimension of pitcher usage. Rest needs to be added to the equation. This is particularly important because the majors and the NCAA differ in this area: While big league starters in a five-day rotation typically have four full days of rest between starts, college starters, by custom, adhere to weekly schedules. This cycle gave rise to a designation for a team’s ace—its “Friday night starter”—so that the No. 1 pitches on the famous day we all love. But this term is colloquial, and to get a more precise picture of the rest afforded to NCAA starters, we need to examine the empirical data.

Days of Rest

We’ll first consider days of rest between any appearance and starts. As there’s no disabled list in college baseball, I omitted rest stints that spanned longer than two weeks, thinking that pitchers were likely hurt in these periods. I’ll again partition the data by class year/season type, and also separate starts by whether 115+ pitches were thrown. That last subset will show whether coaches provided pitchers more rest before pushing them through marathon outings.

NCAA pitchers make more starts on six-plus days of rest than their major league counterparts. However, big league pitchers maintain their customary level (four or five days) more often than NCAA pitchers stay at their supposed weekly rotation cycle. Plenty of NCAA starts come on standard major league rest; the proportion rises from a sizable 22-24 percent of regular season outings to an average of 37 percent come tournament time. Plus, we can see a rising, staircase-like trend in the playoffs bars, showing that the older a college pitcher gets, the less rest he gets. Not only are upperclassmen working harder in starts—they’re resting less for them.

More alarming trends emerge when we focus on the lightly colored short rest category. When it comes to “<115” starts, college pitchers of nearly every class level are coming off zero to three days rest 10–13 percent of the time. Regular season freshmen are an exception. They throw a ridiculous 20 percent of their starts on short rest—although these are surely some of the shorter-stint starts we saw in the first chart.

When we strip out all the shorter starts and focus only on the heavy-usage starts exceeding 115 pitches, we still see ugly proportions. Six to nine percent of the time, NCAA pitchers’ 115-pitch outings come on short rest, a range that stands in both the regular season and tournament. Want to know how high that is? Major league pitchers starting on zero to three days rest never pass the 115-pitch mark. No big league team will use its pitchers this strenuously; college teams do it in one out of every 15 games.

Typically, we think about rest in this “days before” specification. Also very important is the reverse perspective—recovery time after starts. Just as we wonder about the rest leading up to starts of 115+ pitches, we should study pitchers’ recuperation time.

In the regular season, the “after” rest patterns are very similar to the “before” rest patterns. But there are some big changes in tournament play. The age-related trend we saw in the top-right “before” panel is exacerbated in the “after” version, meaning that with every successive class year, pitchers are getting less rest after their starts. More specifically: when it comes to starts under 115 pitches, the rate of short recovery stints rises by about 6.7 percentage points with each class year, swelling to a whopping 30 percent for seniors. Thirty percent! In the bottom-right panel, we see that light-purple percentages following the 115-plus-pitch starts are less consistently age-driven, but are big and unsightly in every instance.

Pitch Smart Guidelines

We’ve explored starters’ per-game workloads and allotted rest. Let’s now put both components together, expand the sample to include all relief outings, and check how well college coaches are following a scientifically-backed standard: the guidelines set out by Pitch Smart, an initiative spearheaded by MLB and USA Baseball to promote safe pitcher usage. Pitch Smart advocates that players aged 19–22 avoid pitching in multiple games on the same day, and stick to the following rest/usage schedule.

These criteria alone don’t provide a full picture of how collegians should be working, as there isn’t an upper limit on a total workload. The guidelines note that the “safe yearly limit varies from pitcher to pitcher,” but it’s impossible for us to pin down a limit without intimate knowledge of each individual hurler. So even though Nathan Bannister’s NCAA-leading 142.1 innings pitched were likely too much, we won’t be able to objectively criticize the University of Arizona for it.

And when it comes to the bullpen, a reliever could throw fewer than 30 pitches every single day of his college career without setting off any Pitch Smart alarm bells. As I analyze Pitch Smart “violations” below, keep in mind that we’re assessing only how many pairs of consecutive appearances can be deemed irresponsible.

We’ll look at the percentage of appearances that would constitute Pitch Smart violations. First, we’ll examine differences at each class level and apply the guidelines to major league pitchers for comparison.

Big leaguers are among the best at sticking to Pitch Smart, picking up violations at a consistent 3.4 percent clip. Lightly used college freshmen are right in line with them, but moving downward, we again see that age trend. With each academic level, NCAA teams do a worse and worse job at following the guidelines tailored for their pitchers. The violation rates blow up in tournament play, when juniors and seniors are committing Pitch Smart infractions in an excessive ~10 percent of games.

Let’s also look at how programs differ in their Pitch Smart violation percentage, separating out probability density curves for NCAA and major league team-seasons.

The most common violation percentage for major league and NCAA clubs is ~3.2. But the average NCAA team is at 4.78 percent, while the average major league team is at 3.36 percent, hinting at the skewing that’s visible in the chart. Reckless usage runs through some teams’ seasons, as NCAA teams violate the Pitch Smart guidelines at upwards of 10 percent, 15 percent, and even 20 and 25 percent. Major league teams hardly eclipse the 6.5 percent mark.

Full Staff Workloads

We also should touch on overall pitching staff management. Do coaches hand off the vast majority of match-ups to a select few starters and relievers, or spread innings out across their staffs?

In answering this question, my tool of measurement will be entropy. I’ve used entropy previously to figure out whether mixing one’s pitches plays a role in predicting ground balls. Here, it’ll measure usage concentration. If a coach were to concentrate his match-ups in the most extreme way possible—giving them all to a single pitcher—that team-season would register an entropy of zero. If workloads are more equally-split and spread across more staff members, the higher the entropy. Larger entropies indicate balanced usage and a well-managed staff.

First, let’s consider the probability density curves of pitching staff entropy for major league and NCAA team-seasons.

We see a clear dichotomy between the distributions: Major league entropies are predominantly better than NCAA entropies. One caveat here is that college teams don’t operate with the same resources at their disposal. They can’t fill spots with trade acquisitions, top prospects or freely available replacement-level players. NCAA teams must work with the pitchers already on their staffs. But can NCAA programs do better to spread out innings? I think so, as the overlap between the probability density functions shows that it’s feasible for an NCAA team to allot batters faced like a major league team.

Which college programs are already doing that? And which manage their pitching staffs with minimal balance? Let’s get an idea of both ends of the entropy spectrum with the table below.

Florida and UCLA are two esteemed programs that each had three strong years and one below-.500 year in the 2012–2015 period. When it comes to staff management, Florida ranks among the most balanced, while UCLA is the most concentrated. (Both results are unsurprising.) The two schools do well to demonstrate what these entropy figures mean.

UCLA coach John Savage entrusted just a few hurlers with the lion’s share of innings. On average, 54.1 percent of UCLA match-ups went to the three hardest-working Bruins pitchers; in Florida, the three hardest-working Gators threw to 39.3 percent of the club’s opposing batters. If we compare the top five workhorses from both teams, UCLA’s groups pitched to 77.9 percent of all batters, while the Florida squads threw to 57 percent. Florida was even better than the median major league team, whose three- and five-player percentages were 41.3 percent and 58.2 percent, respectively.

It’s worth noting that Florida had the luxury of more talented pitching staffs, allowing coach Kevin O’Sullivan to spread out innings and worry less about blowing potential victories. From 2012–2016, 25 pitchers were drafted out of Florida, versus 15 for UCLA. Even more indicative of quality, just two Bruins hurlers were taken in the first 150 picks of the past five drafts, compared to nine for the Gators—a tally that includes A.J. Puk, ranked by Baseball America as the No. 1 talent in the 2016 draft. Coach Savage did what was necessary to win with fewer horses in his stable. But he and the other low-entropy programs did pitchers no favors in the process.

Tommy John Surgeries

Following all these results about usage is the question of whether collegians are incurring more injuries. To that end, we’ll consult Jon Roegele’s list of Tommy John surgeries, and combine it with the Chadwick Register and Steamer’s archived projections. Altogether, this allowed me to figure out the Tommy John rates for pro pitchers who either attended college or signed as teenage amateurs (domestic or international). In alignment with the utilized college data, I targeted pitchers who were born in 1991 and later.

The results don’t come out in the NCAA’s favor. Among collegians who went pro, 7.9 percent had the UCL reconstruction surgery. Meanwhile, the surgery has been needed for 6.8 percent of teenage amateur signees. Now, this result does come with a caveat: it’s not a significant difference. Our confidence level that this +1.1 percentage point differential is real is just 5.5 percent. More data and testing are needed to know whether college pitchers actually do get hurt more frequently.

Final Remarks

College pitchers shouldn’t throw such lengthy outings, they shouldn’t so often violate Pitch Smart guidelines, and their total workloads should be lightened. Since college coaches’ prime duty is to win, they have little incentive to change on their own. I believe that limits should be imposed across the college game. Making the Pitch Smart guidelines mandatory would be a start. Such restrictions may rankle coaches, but the limits would be fair for all of the country’s D1 programs.

Who can rein college coaches in? One wonders if MLB and the MLBPA could aim to get involved, given their support for the new high school pitching restrictions. Extreme amateur usage should be a bipartisan issue for the league and union, since injuries spoil teams’ investments and may lead to smaller major league player contracts. Perhaps MLB and the MLBPA could tie the issue together with additional scholarship funding.

But before MLB and the union even consider intervening, the NCAA should be pushed to live up to its own credo. College sports’ powerful overseer says it is dedicated to “safeguarding the well-being of student-athletes.” If that’s really the case, the NCAA should protect its pitchers by imposing common-sense usage restrictions.

References & Resources

• All college game log data come from Bryan Cole’s Github page, which was originally forked from Meredith Wills’ Github page
• All major league data come from PITCHf/x
• Tommy John calculations were created with Jon Roegele’s list, Ted Turocy’s Chadwick Register, and Steamer’s archived projections
• Baseball-Reference’s draft pages
• Google doc showing days of rest breakdowns, as well as school rankings in entropy and Pitch Smart violation percentage
• Keith Law’s Twitter feed, which is full of notes and observations about college pitcher abuse
• Pitch Smart guidelines and risk factors
• Stack Exchange, “Can a probability distribution value exceeding 1 be OK?”
• Matthew Leach, MLB.com, “Fleisig: Fatigue key factor in Tommy John injuries”