THE BOOK--Playing The Percentages In Baseball

"Essentially, there is no reason to believe that the walk rate plays any predictable role in the future development of a young player.”

Corey Patterson might disagree.

I haven’t done a study, but I’ve followed the minor leagues very closely for several years. I don’t pay for Bill’s site anymore so I can’t read his stuff, but I assume he did a study of the topic.

Yes, if one has a low walk rate there is more room for them to grow as a player. But I would argue that those with low walk rates often don’t see much improvement in their PD as they advance. But that is something I would like to look into further.

One player who is highly debated currently is Josh Vitters of the Cubs. He was drafted very high out of HS and doesn’t have a strong walk rate. In fact, he barely walks at all. Some keep him very high on lists because he has shown great contact ability, others have dropped him lower due to concerns with how he projects to hit against better pitching. Those who have dropped him down on lists (including myself) see a player who will too often find himself behind in the count.

James makes an interesting point. But I think the bigger question is “how do players develop their walk rate.” We know it increases but there needs to be a line somewhere. How does a sub 3% walk rate (Vitters) progress?

Having more room to grow isn’t necessarily a good thing. It seems likely to me that by the time those guys to develop the plate discipline necessary to be good players, their other skills have deteriorated too much for the increased walks to be significant. Or, possibly, guys (like Patterson, who was mentioned here) exhaust their chances in the majors because they take too long to learn how to take a walk. Patterson might have amazing plate discipline when he’s 35, but I don’t see him getting many chances to prove it because he’s been so bad for so long.

JD/2: I agree. There must be some sort of cutoff. I agree with you on Vitters—I get that there’s “room to grow” for him in his plate discipline but how often does it happen? I can’t read Bill’s study either so maybe he addresses it. I’d like to see how often a player with a 3-5% BB rate in Low-A ball develops into a much better BB Rate—or, how often someone with that low of a BB rate at a low level even becomes an MLB regular.

Obviously, the results or conclusions coming out of this study depends on your definition of “develop” (and, BTW, that is one thing that bothers me about some of James’ studies - the way he throws out words that have no exact definition).

More importantly, though, you cannot talk about any kind of “development” in terms of true talent by looking at the changes between a selectively sampled set of data and one that is randomly selected.

The most obvious example is if I do a “study” whereby I say I want to look at the difference in “development” between players who hit very low HR (per PA) and ones who hit very high HR.  Obviously if I look at any future time period, either in the near or in the far, both groups will regress towards the mean for all players, and that would have very little if anything to do with their “development” per se.

James is doing essentially the same thing.  He is looking at low and high walk players early in their careers and seeing how each group “develops,” by comparing those high and low walk seasons (among other data points) to future performance.  Well, we KNOW that regardless of how their true talent “develops” that the low walk group will regress higher and look like they are “developing” into better walkers and the high walk group will regress lower (assuming they were higher than the mean for the age in the first place), looking like they are not “developing” as walkers.  In reality, of course, once you factor in normal aging, the low walk group probably increases their walks a lot and the high walk group probably stays around the same, even with regression towards the mean.

So, for example, if James finds no significant difference in terms of their walk rate development, once you factor out the regression toward the mean, that would actually suggest that the low walk group did not develop as much as the high walk group, since you should see a much more marked increase in their walk rate, due to regression towards the mean alone.

As far as the other stats that James is looking at in terms of “development”, you would have to adjust them for regression towards the mean as well, unless both groups had around the same levels in their “other stats,” in which case both groups would have around the same amount of regression and you could essentially ignore it when comparing the two groups.

There are two ways to do this kind of study to avoid this problem.  One is to do the regression or adjust for it in the analysis.  The other way is to use an “out of sample” data as your first comparison point, so that it is random and not selectively sampled and therefore represents a true talent sample.  So, for example, James would use season X data to create the two groups, low and high walk players, and then use season X-1 (or all prior seasons) as the initial comparison point, as season X-1 will already include the regressions. He could also use season X+1, although that would obviously start to include some “development” already.