Talking Sabermetrics: Pitchers' Control and Injuries

We have mentioned it a number of times. During Jeff Luhnow's first season, the Astros have shown a preference for acquiring strike-throwing pitchers. This is apparent among the drafted college pitchers, as well as many of the minor league pitchers acquired in trades at mid-season.

There might be any number of reasons. The most obvious is the obvious. If you looked at the typical pitching coach's list of commandments, you will probably see "throw strikes" at, or near, the top of the list. If we assume that control pitchers also are most likely to exhibit command within the strike zone, then we can add "location, location, location" to the pitching admonitions. We can also speculate that the preference for control pitchers may be exploiting a market inefficiency. Because strike outs are seen by teams' scouts as an indication of "stuff," possibly control has become undervalued relative to the other main peripheral statistic, strike outs.

Fangraphs' Jeff Zimmerman raises another possible attribute of pitchers' control: Does control and command of the strike zone tell us something about pitchers' future injury risks? His article "Injury Chances for Strike Throwers" inspired today's Talking Sabermetrics topic.

The FG article starts with an interesting quote from Tim Kirkjian of ESPN, discussing Oakland GM Billy Beane's view regarding the traits he wants in young pitchers:

GM Billy Beane doesn’t require power, he wants outs without walks. Plus strike throwers generally have good mechanics that help prevent injury. Beane also isn’t afraid to go with young pitchers, what at least in theory are less likely than older ones to get injured.

Beyond the obvious reasons to favor strike throwing pitchers, a sabermetric oriented GM believes that control pitchers have a lower risk of injury. Certainly the proposition has some logic. Control pitchers must develop good, repeatable mechanics; and good mechanics should be associated with a reduced injury rate.

I can think of two more similar explanations for believing that control is related to injury rates. First, over-throwing is the enemy of control, and presumably over-throwing increases the chance of arm injury. Second, we would expect high velocity to be correlated with reduced control--maybe a better way to state this is that lessor control can be tolerated from hard throwers. While not all hard throwing pitchers are injury prone, generally throwing at a very high velocity puts more stress on the arm joints.

The explanations may be logical but that doesn't prove that the connection between control and proneness to injury is true. Zimmerman tried to test the theory by comparing various measures of control (strike %, zone %, non-intentional BB%) with pitcher DL visits. Zimmerman's analysis produced mixed / weak results to support the notion that control pitchers are more likely to be reduced injury risk. Only pitch f/x zone percentage seems to be related to a lower risk of injury. Of course, it's possible that DL visits are too imprecise a measure of pitcher injury risk, since it isn't limited to arm injuries nor does it reflect the severity of the injury.

But that's not the end of the story. Zimmerman notes that the data supports the idea that pitchers with really bad control are more likely to be injury risks:

While I was only able to find one of the three categories that showed strike-throwers being healthy, I did find that extreme non-strike throwers had a higher likelihood of ending up on the disabled list.

This finding also raises a cause-effect issue. Perhaps the extreme non-strike throwers are already suffering injury impacts while they pitch, and this leads to poor control. Or maybe the lack of control puts the pitchers in too many high pressure, high pitch count innings, which leads to injury. However, even if the finding is due to these alternate hypotheses, it is still relevant to front office pitcher acquisition decisions.

Zimmerman's article uses the following thresholds for extreme non-strike throwers: <60% strikes; <47% zone; >10% non-intentional BBs. Pitchers in these ranges experienced a disabled list rate of 45% - 49%, roughly 10% - 15% higher than normal.

One of the best templates for a "Talking Sabermetrics" article is to find a research article at Fangraphs or Beyond the Boxscore, and then apply the findings to Astros' players. If you have followed the Talking Sabermetric series, I suspect that you already noticed the cookbook approach. That's what I did here.

I examined the status of Astros' starting pitchers (excluding those whom are non longer with the organization) relative to the three "non-strike thrower" thresholds. (I used BB% instead of non-intentional BB%, by the way.) I calculated strike percentage from TexasLeaguers.com pitch f/x data, with the remaining two categories derived from Fangraphs. The results are shown below.

Setting aside whether these measures help us evaluate injury risk, they obviously tell us a lot about each pitchers' control. Fortunately, neither Harrell, Norris, or Lyles are among the extreme non-strike throwers in any category. All three pitchers are expected to be part of the Astros' rotation next year, and they didn't have the control problems that would mark them as particular injury risks. On the other hand, based on the Fangraphs' data, their control results don't suggest that they have an above average chance of avoiding injury (a zone percentage above 51.5%).

Edgar Gonzalez, Dallas Keuchel, and Kyle Weiland each qualifies as an extreme non-strike thrower in at least one category. Gonzalez generally appears to have good control (lowest BB rate among these pitchers). However, he fell below the 60% strike threshold. Notably, strike % is the weakest of the three indicators for injury risk. So, I don't think the result is alarming for Gonzalez (particularly given the sample size).

I was interested in including Weiland in the comparison, because we know that he pitched with a shoulder injury during his stint on the major league roster. And, indeed, he he had the lowest zone percentage among the selected Astros' starters. The only Astros' starting pitcher with a lower zone percentage was Gallaraga, who was released before the season ended.

Dallas Kuechel failed in two categories (zone percentage and BB percentage) and barely met the 60% threshold for strike percent. Keuchel's results seem more alarming; his control was terrible. Perhaps even more concerning is that Keuchel's minor league record indicates that he has succeeded as a control pitcher. His major league walk rate in 2012 is more than twice his minor league career rate. During the season, I questioned whether Keuchel had changed his approach to pitching when he was called up to the majors. However, given the results in the fangraphs article, I now wonder if injury or fatigue could have been involved in his control issues. He isn't a pitcher I would expect to have injury problems, but a medical check might not be a bad idea.

Do you think that the possible connection between injury risk and control tendencies will affect the Astros' off-season pitching moves?