College pitcher development patternsby Jeff Sackmann
March 04, 2010
Last week, I presented some research on the aging patterns of college hitters. Of course, pitchers develop too, and it's interesting to see how they do so at the college level.
If you want some background on the project, I encourage you to click over and read last week's article. This week, I'll cut straight to the chase.
To get us started, let's look at some multipliers for a relatively select group of players. To zero in on a pool with a lot of draft-ready talent and avoid some potential pitfalls, we'll start with those pitchers who mostly served as starters and faced at least 300 batters in two consecutive years:
Pair Players $HR $K $W $HP $BABIP Fr -> So 115 1.12 1.04 0.99 0.99 1.03 So -> Jr 223 1.11 1.05 0.99 1.02 1.03 Jr -> Sr 226 1.08 1.03 1.01 0.98 1.02Each number in the table is a multiplier showing how much a rate changed from one year to the next. (The rates are calculated as outcomes per batter faced.) For example, it shows that the average sophomore gave up 12 percent more home runs per batter faced than he did the previous year.
... say what?!
In fact, it isn't just the apparent increase in home run rate that raises eyebrows. The moderate increase in strikeouts each year is to be expected, but the changes in walks and hit batsmen are barely enough to merit the word "changes," and while we might not anticipate a change in BABIP, a slight decrease would make more sense from pitchers who are (ideally, anyway) learning to coax more ground balls.
As I did with batters, let's see what happens if we narrow the field down to those pitchers from the eight most competitive Division One conferences:
Pair Players $HR $K $W $HP $BABIP Fr -> So 35 1.15 1.03 1.03 0.96 1.02 So -> Jr 78 1.12 1.04 0.97 1.00 1.02 Jr -> Sr 82 1.09 1.05 1.00 0.95 1.01With under 100 pitchers representing each pair of classes, there are some sample size concerns right off the bat. For the most part, though, these are the same results for all of Division One. I tinkered around with the parameters quite a bit, including and excluding relievers and varying the playing time minimums. No matter how you slice it, the results look roughly the same.
If we're going to explain these counterintuitive findings, we're going to have to look elsewhere.
Better pitchers, worse performanceIn essence, the problem is simple. We assume that as players progress through college, gaining physical size, fitness, experience, and coaching, they improve. We certainly saw that in the results last week for batters. But here, we're seeing numbers that suggest that performance is about the same from year to year, if not worse.
Here are a few ideas that may explain it:
- As pitchers age, they learn to throw harder, and faster pitches, combined with aluminum bats, result in more power. This doesn't hold true at the major league level, but when aluminum is involved, you never know. This theory is appealing because it could explain the increase in both home run rate and BABIP. Unfortunately, it leaves us saying, "Aluminum bats are weird, huh?" We're unable to test the theory with any available data that I know of.
- Some pitchers get overworked as underclassmen, resulting in lessened effectiveness the following year. This is a possibility, though I think that more and more coaches are aware of how to manage the workloads of young pitchers. The parameters of this study also exclude those pitchers who are overworked and then get hurt, thus missing the playing time cutoff of 300 batters faced.
- As college pitchers post better results, they are moved into more prominent roles. Many freshmen are used as midweek starters and occasional relievers, while established pitchers start the higher-profile weekend games. This is intuitive, and best of all, it is testable.
Everybody's working for the weekendEspecially in top conferences, the difference between weekend competition and midweek competition is substantial. After the first few weeks of the season, teams typically play three-game sets against conference opponents on the weekend and play a game or two against non-conference rivals on Tuesday and Wednesday.
If you're playing in the SEC, one of the toughest conferences in college baseball, it stands to reason that weekend games are tougher than midweek games. As we'll see, the difference can be striking.
With enough data on every pitcher, we can take "strength of schedule" (SOS) to a whole new level. Since hitters play every day, the specific mix of teams that a batter faces doesn't matter much—it's very close to the mix that the team faces. But for pitchers, it's a different story. So instead of using the team's overall SOS to adjust a pitcher's stats, it's more accurate to use a pitcher-specific SOS.
For example, here are SOS ratings for the mix of opponents faced by six different Louisiana State pitchers last year:
Pitcher SOS Louis Coleman 1.40 Anthony Ranaudo 1.38 Austin Ross 1.31 Chris Matulis 1.17 Ben Alsup 1.12 Jordan Nicholson 1.07Any guesses who started weekend games for LSU last year? These are huge differences. No team had an SOS as high as 1.31, while 1.07 is fairly common. Coleman and Nicholson may have pitched for the same school, but they were presented with very, very different challenges.
While many top prospects (like UCLA's Gerrit Cole) spend their whole college career in the weekend rotation and thus don't see the difficulty of their competition change much, that isn't true for the "average" college starter. For instance, Blue Jays prospect Chad Jenkins had a 1.06 SOS in 2008 and a 1.15 SOS a year later. Matt Bashore went from 1.01 to 1.07. Diamondbacks farmhand Jordan Meaker went from 1.00 to 1.19 between '07 and '08.
The proof, or at least some puddingFor the pitchers in elite conferences, I re-ran the study using pitcher-specific opponent ratings, rather than team-wide numbers. Thus, if a pitcher's numbers got worse simply because he faced tougher competition, that effect should be neutralized.
Does adjusting each for each pitcher's competition make a difference? Well, it doesn't solve the paradox of the rising home run rate, that's for sure:
Pair Players $HR $K $W $HP $BABIP $SOS Fr -> So 35 1.13 1.04 1.01 0.95 1.01 1.01 So -> Jr 78 1.12 1.05 0.97 1.00 1.02 1.01 Jr -> Sr 82 1.09 1.04 1.00 0.95 1.01 0.99The results are about the same. They are a tiny bit more appealing than the last set of numbers, but only a tiny bit.
As noted in the "$SOS" column, the average starter doesn't see his competition change that much from year to year. I assumed that a large number of hurlers moved from midweek gigs to weekend assignments. That might be true, but perhaps many of the midweek starters, in their roles as swingmen, don't get enough innings to show up in these results.
For whatever reason, the counterintuitive conclusion remains. As college starters develop, they give up more home runs, even adjusting for the specific opponents they face. As if college baseball didn't hold enough mysteries already, we can add one more to the list.
References and Resources
Thanks to John Barten and Josh Kalk for their help.
Jeff Sackmann is the creator of MinorLeagueSplits.com. With Kent Bonham, he founded CollegeSplits.com. Jeff and Kent blog about college baseball and the draft, and you can follow them on Twitter for bite-sized snacks of minor league and college stats. Jeff also has an email address.