Avoiding the hook

Almost three decades ago, Bill James came to the important conclusion that high-strikeout pitchers last much longer in their careers than do comparable low-strikeout hurlers. Ever since, teams and fans familiar with James’ work have subscribed to the belief that pitchers who strike out a lot of batters will last longer, and are therefore more deserving of long contracts, than pitchers who don’t.

The problem is, it’s not clear that James was right. As we found last week, in grouping pitchers by wins, James made a serious mistake. Although a randomly selected group of pitchers who win a given number of games in a season should be equal to another randomly selected group of pitchers with the same win totals, grouping pitchers based on their strikeout rates is most certainly not random, and thus it creates all kinds of potential issues.

And as we found last week, those issues are important. As a general rule, pitchers who strike out a lot of batters are better than those who don’t; a corollary is that a high-strikeout pitcher with the same number of wins as a low-strikeout pitcher has probably been somewhat unlucky, with the opposite being true for the low-strikeout guy.

The ultimate consequence is something James would call “quality leakage”—the high-strikeout pitchers we flag are inherently better than their low-strikeout counterparts, and so of course they end up lasting longer. Once we correct for that issue, we find no statistical difference in the lasting durability of power and finesse pitchers.

That column inspired a lot of interesting reader mail. Today I’d like to focus on two of the more interesting suggestions. The first comes from Jake, who wonders:

After I read your power vs. non-power pitcher career analysis, I started thinking about how, knuckleballers and Jamie Moyer aside, a bunch of those who pitched successfully into their 40s are not just high-K pitchers, but *big* high-K pitchers. Randy Johnson: 6-foot-10. Nolan Ryan: just ask poor Robin Ventura. Roger Clemens: Even modulo the alleged chemical assistance, he’s a pretty big guy.

On the other hand, the two little guys that come to mind with huge strikeout figures are Koufax and Pedro, and both have had arm troubles. (The only difference between Koufax and Pedro’s careers is that nowadays we know not to let anyone throw 176 pitches in a game, so we’ve been able to save Pedro for a few more years.)

(The only real counterexample I can think of offhand is Jamie Moyer…I have no explanation for him. Greg Maddux and Tom Glavine are a year or two away from being counterexamples.)

So I am wondering if “high strikeout rate” and “longevity” are more correlation than causation: To me, it seems more intuitive that 6-foot-10 Randy Johnson can throw 98 mph, than that 5-foot-11 Pedro can—so if it’s easier for the big guy, he can continue to do it longer.

Two years ago, when I looked at how pitchers of different sizes perform, I found that tall pitchers have only a marginally higher strikeout rate than do short ones. Nevertheless, Jake’s theory struck me as worth checking out. After all, it is somewhat intuitive that bigger guys would have stronger, more resilient bodies.

To test this theory, I used much the same method as I did last week: first, dividing all pitchers since 1946 with at least 500 innings pitched before the age of 26 into three groups based on their height, and then comparing the shortest and tallest thirds.

In this case, there was not as much quality leakage as when we split up the groups by strikeout rate, but I did have to remove a few pitchers from each sample to even things up. In the end, we had 112 short pitchers and 112 tall pitchers, and each group had the following average line in their career before turning 26:

Category     IP     FIP
Tall         777    4.30
Short        782    4.36

FIP, of course, is fielding-independent ERA, which we are using for reasons explained last week, and for which a definition can be found here.

Clearly, the two groups are very similar at a young age. But what happens as they get older? Here are their stats after turning 26:

Category     IP     FIP
Tall        1028    4.40
Short       1077    4.35

In all, the tall pitchers see a 32% jump in innings pitched and a 2% rise in FIP, whereas the short pitchers throw 38% more innings with essentially zero change in effectiveness. The differences are small and insignificant, and intuitive though the theory might be, it definitely seems to be one that we can reject. There is no evidence that tall pitchers last longer than short ones.

But we’re not done, because reader Mark Heil came through with another very intriguing idea:

If Bill James is truly interested in figuring out what leads to long term effectiveness, I’ll bet that walks/IP will be a better measure. Pitchers who walk one batter per every three innings or fewer last longer. Schilling, Maddux, Carlton, Seaver, Gibson, Moyer, Radke, etc. all had outstanding control. Pitchers like Nolan Ryan and Randy Johnson developed better control later in their careers. There will be some exceptions—sinkerball pitchers like Carlos Zambrano come to mind. In my biased opinion, the quality of a pitcher can best be measured by his walks to innings pitched.

Walks are sort of the flip side of strikeouts, but I don’t think that Mark’s theory is any less reasonable than James’, so why not test it? Again, we’ll split our sample into three groups—this time based on walk rate—and select a high- and low-walk group.

After paring down the sample to make the two groups of equal quality, we end up with 92 pitchers in each. The high-walk pitchers averaged 4.2 walks per nine innings before turning 26, while the low-walk pitchers averaged just 2.6. (As with every other statistic in this article, all those numbers have been adjusted for league and year.)

Again, the two groups are very similar before age 26:

Category     IP     FIP
Low-walk     740    4.34
High-walk    740    4.41

But the question, once more, is what happens after they turn 26? The answer is very interesting:

Category     IP     FIP
Low-walk    1113    4.36
High-walk    890    4.42

While both groups see their FIP barely move, the low-walk group pitches way more innings after age 26 than does the high-walk group—an increase of 50 percent for the low-walkers versus just 20% for the high-walkers! Statistically, that difference is fairly significant—for you stat wonks, the t-value is 1.8.

And of course, it’s easy to see how big that difference is in practical terms. If you have two equally valuable 26-year-old pitchers, but one walks relatively few batters while the other walks many, you would expect 25 percent greater value from the pitcher with good control than from his counterpart with poor control!

Now, to really evaluate the situation, we would have to try to learn why young pitchers with low walk rates have longer careers than do young pitchers with high walk rates but equal effectiveness. The difference in longevity might have something to do with control. Or it could be a symptom of lower pitch counts. Or it could be something else. That’s the next question we’ll have to answer.