All about college walk ratesby Jeff Sackmann
May 06, 2010
Take a look at the Division One standings for adjusted OPS, and you might be surprised by what you find. These numbers are tweaked to take park and schedule into account, so at least in theory, we should be looking at the guys who are having the best seasons.
There are some big names on display: Rob Segedin, Yasmani Grandal, Kyle Parker, Michael Choice and Anthony Rendon are among the top 15. But so are Chad Salem (Manhattan), Pat Biserta (Rutgers), A.J. Kirby-Jones (Tennessee Tech) and Kevin Tokarski (Illinois State).
That latter group is full of talented players, sure, but are they really in the same league as Grandal and Rendon? Any scout would say no, and common sense chimes in with a negative as well. So what's going on?
Part of the explanation is that, given 300 teams, you're going to get a couple of guys who have otherworldly seasons, whether their talent dictates it or not. That probably accounts for at least a couple of the surprising names. Also possible is that the park and schedule adjustments aren't quite right. (In the case of Biserta, I recently took a closer look, and at least in that case, the multipliers appear to be solid.)
Neither of those explanations is particularly satisfying. Can we do better?
The case of Michael Choice
Choice, a center fielder at Texas-Arlington, is an interesting case study. UTA plays a tougher-than-average schedule, but it is definitely a second-tier school, facing competition that doesn't compare with, say, the SEC or the Pac-10.
Choice is the best hitter UTA has seen in a long, long time. For example, he set the school's single-season record for intentional walks three weeks ago. Whether intentional or not, he's getting a whole lot of free passes, with 59 walks in 217 plate appearances through last weekend. One quarter of those are intentional, but that still makes him look pretty good in the pitch-selection department.
Needless to say, that walk rate of 27.1 percent isn't going to hold up in the pros. Even his unintentional walk rate of 21.8 percent is unlikely to reflect his true talent level. Last year's major league leader was Adrian Gonzalez, at 17.5 percent.
When a player gets so many intentional walks, it seems reasonable to assume that many of his unintentional walks are really somewhere in between. All the evidence points in that direction for Choice. His unadjusted 1.328 OPS is the only one on the team above .900. Preston Beck, the guy who usually hits cleanup behind Choice, isn't even slugging .400. If you were coaching Stephen F. Austin next weekend, wouldn't you tell your pitchers to keep the ball away from Choice's bat?
Should walk rate adjustments be non-linear?
Enter dead horse, stage left. Recently, I've presented some research showing how college players perform against elite competition—specifically, drafted players. My findings suggested that performance against draftees is no more predictive than properly adjusted results against the general population of college players.
In short, it looks like the relationship between college and the pros (the lower rungs of the pros, at least) is linear. That is, there's no invisible wall that consistently stops certain types of players from progressing, given that they've already had some success at the college level.
What if walk rate is an exception?
Choice's results this year are just an anecdote, of course, but they suggest why this would be the case. Most stats reflect, at some level, a player's skill. For a batter, avoiding strikeouts is a skill. Hitting home runs (or, at least, a lot of hard-hit balls in the air) is a skill. In some environments, taking a walk is a skill. But not always.
Taking an intentional walk, obviously, is not a skill. For a pitcher, not issuing intentional walks is not a skill. For the issuing team, intentional walks are a strategy. That's why, for many types of analysis, we leave out intentional passes when counting walks.
But it isn't that simple. As anyone who ever watched Barry Bonds at the plate with Bengie Molina on deck can tell you, unintentional walks are often a strategy as well. Bonds had a great eye, of course, but it would miss the point to ascribe all of his unintentional walks to his pitch-selection skills.
The point is this: Strategy is non-linear. If Prince Fielder gets hurt and Ryan Braun's walk rate goes up, is it because his pitch selection suddenly improves? Of course not. But unless we include a "strength of surrounding lineup" adjustment, no stats are going to explain that to us.
If we use the increased walk rate to predict Braun's performance when Fielder returns to the lineup, we're likely to be wrong. When dealing with major leaguers, we know what's going on—if you're deciding whether to put a certain player in your fantasy lineup, you have some awareness of where he hits in the lineup, and how it might affect something like this. But when analyzing thousands of college hitters, such details get lost in the shuffle.
So we find ourselves in a tricky situation. For your average college hitter, walks reflect skill. For anyone, intentional walks are all but meaningless. For an elite college hitter in the middle of a non-elite lineup, unintentional walks, some of which are probably somewhat intentional, reflect ... well, what, exactly?
Testing the hypothesis
Let's make some assumptions more concrete. If a draft-worthy hitter is playing for a non-elite school, he probably outclasses the rest of his lineup. Thus, he won't see very many pitches to hit, and he'll rack up a walk rate that, even adjusted for level of competition, overstates his talent level.
Sound plausible? It does to me.
Defining "elite" as a school's membership in one of the top eight Division One conferences, I started by comparing walk rates in the college season to walk rates in the Cape Cod League season. Most of the players from non-elite schools had to be pretty good to get into the Cape, and presumably they didn't stand out so much among the higher-quality lineups on display there.
So, if guys like Choice and Kirby-Jones are getting cheap walks during the regular season that they wouldn't get if they hit in the middle of more evenly balanced lineups, we'd expect them to lose some of those walks in the Cape.
No dice. Walk rates in the Cape are almost identical to walk rates during the D-1 season. That applies equally to players from elite schools and players from non-elite schools. The similarity is eerie—the difference is less than 1 percent.
What about players who go on to play in the pros? That's what we really care about, right?
And finally—finally!—we have something. I used the pool of 900 or so players who, drafted or signed in the last three years, have logged plate appearances in both D-1 and the minors, then compared their college walk rates with their results from their first two seasons in the pros. As it turns out, about half of those players came from elite conferences.
In general, walk rate for these players is a bit higher in college—no surprise there. In D-1, the BB/PA (adjusted for level of competition) is 9.8 percent, compared with 9.3 percent in the pros. But check out what happens if we divide the pool into players drafted from elite and non-elite schools:
BB/PA College Pros Elite 9.8% 10.0% Non-elite 9.7% 9.1%That isn't the difference between a star and a washout, but it's enough to make you sit up and take notice. Your average guy from Rice, Vanderbilt or Stanford is likely to see his walk rate hold up in the low minors. That's plausible, as he'll be facing high schoolers after doing battle for three years with polished college arms. But players from schools like UT-Arlington see their walk rates take a hit.
The Choice is yours
(You didn't really think I was going to avoid a "choice" pun for the entire article, did you?)
Back to our favorite small-college first-rounder. To evaluate his pro-worthiness, it looks like we need to adjust his walk rate one more time. Already, we're committed to ignoring the intentional walks, which cuts his total of free passes by a quarter. We then do a conventional schedule adjustment, which pushes him up by a walk or two—UTA doesn't play a tough schedule, but their competition is a bit more challenging than that of the average D-1 program.
But, if we're looking at his walk rate in terms of what we can expect in the pros, we need to knock off about 10 percent of what's left. So instead of the eye-popping 59 walks in 217 plate appearances (27.1 percent), we've got something like 41 walks in 202 plate appearances (20.3 percent).
If scouts have watched opposing teams consistently and obviously pitch around Choice, maybe we would want to knock that number down a bit more—there's no reason an algorithm designed for 300 college teams should take precedence over all other information. But no major league club has enough scouts to make that kind of judgment about every small-college semi-prospect.
Now we have some numbers to back it up. Not just the implication that walk rate adjustments are non-linear, but also a stab about what those non-linear adjustments should look like. Every once in a while, beating a dead horse pays off.
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.