Tom M Tango, along with co-authors Mitchel Lichtman and Andy Dolphin, have written a superb book of in-game strategic analysis called The Book. Tom has graciously given us permission to publish the following excerpt.
In discussing strategies of intentional walks, we tend to focus on the “yes/no” question: should the pitcher walk the hitter, or pitch normally to him? In reality, there are intermediate answers as well, as a pitcher need not throw pitches down the middle of the strike zone, or for that matter within the strike zone at all. In other words, the pitcher can pitch in such a way that the odds of a non-intentional walk are higher, while the odds of a well-hit ball (should the batter not be walked) are lower. While we’d love to be able to conduct a thorough analysis of this, the available data do not include accurate pitch locations.
Instead, we will have to use certain situations as a proxy for what we think a pitcher is likely to do in a given situation. In other words, if the situation allows for an intentional walk (good hitter at the plate, poor hitter on deck, one or two outs, batting team leading or tied, and first base is open) but the pitcher opts to pitch, then we would guess that the pitcher is probably at least being a bit more careful than he would otherwise be. On the other hand, if the situation were identical but a good hitter was on deck, we assume the pitcher is pitching normally. If our theory were correct, in the first situation we would see more walks and fewer well-hit balls; in the latter we would see the reverse. Is this what we actually find?
From 2000–2004, there were over 800,000 plate appearances in which the batter had over 200 PA during the five-year span and in which the next hitter was not a pinch hitter. In these plate appearances, an intentional walk was issued 0.7% of the time. We next define a “likely intentional walk situation” as one in which there are one or two outs, the game is tied or the pitching team is trailing, first base is open, and there is a man on second or third. And indeed, 84% of intentional walks occur in these situations, at a rate of one intentional walk in every twelve plate appearances.
Now let’s see how well good hitters perform in these situations when they have protection and when they don’t. We define “protected” in this instance as having an established good hitter on deck, and “unprotected” as having an established poor hitter on deck. The following table lists the average stats of these good hitters when protected and when not protected, with both stat lines scaled to 600 plate appearances. For comparison, we also computed the average stats when the pitching team is ahead by at least two runs (again with a good hitter at the plate), a situation that is never an intentional walk situation.
So, no surprises. With the weak hitter on deck, we get considerably more walks and a few more strikeouts, both of which we expect to see if the pitcher is pitching around the corners. Specifically the walk rate goes from 13.8% with protection to 15.2% without protection, while the strikeout rate increases from 17.3% to 18.0%. And, regardless of the on-deck hitter, the pitcher is avoiding the batter more than he would in a poor walk situation—when ahead, the walk rate is 12.0% and the strikeout rate is 16.1%. So, we can reasonably conclude that we are measuring what we expected—that pitchers throw more balls when a walk isn’t such a bad prospect, thus dramatically increasing the number of walks and also somewhat increasing the number of strikeouts.
However, not everything in this table makes sense. The entire point of protecting a batter is to improve his offensive output (wOBA) by forcing the opposing pitcher to pitch to him. And indeed, we saw above that opposing pitchers pitch to protected hitters, something that is evidenced by the fewer walks. However, when the ball is put into play, we see no significant difference between how the two sets of hitters perform. The unprotected hitters have a wOBA of .395 (counting only balls that are hit), compared with .391 for protected hitters. The difference of .004 is not statistically significant. For comparison, the good hitters in the “leading” situation have a “contact” wOBA of .404, which is a somewhat statistically significant deviation from the other values.
In short, protecting a star hitter appears to accomplish very little. He indeed gets fewer walks; however, there is no evidence that he gets more hittable pitches, since the pitcher always avoids pitching to a good hitter when the situation would call for an intentional walk.
Where we do find a significant difference in the pitcher’s approach is in situations in which a walk would or would not be considered. The following table contains the average statistics (again, from 2000–2004) of all hitters, split between situations in which an intentional walk would be considered and those in which one would not be considered. Because defensive positioning adjustments due to baserunners can change the batter’s wOBA (as we discuss in the chapter on baserunning), we consider only at-bats in which first base is open and second and/or third are occupied. If the pitching team is ahead or there are no outs, we call it a non-walk situation; otherwise we classify it as a potential IBB situation.
While before, we were trying to draw conclusions from a few thousand PA, we now have over 20,000 to produce more accurate results. So not surprisingly, we find once again that, in situations in which a walk is a good option, the pitchers indeed walk the batter more frequently. Also, by pitching around the corners, the batters occasionally will swing and strike out, leading to more strikeouts.
However, what we hoped to find was that, when pitchers pitch around the corners, batters tend to make worse contact (if they make contact at all). And we don’t see this. Instead, we again find that, if walks are ignored, the two types of hitters perform equivalently in wOBA. (The slight increase in strikeouts is compensated by a slight increase in the fraction of balls hit well when contact is made.) In short, we simply cannot find any evidence to suggest that the pitcher’s approach has any significant impact on the batter’s stats, aside from the obvious changes in walks and strikeouts.
In terms of an alternative to intentional walks, pitching around a hitter accomplishes very little, as it merely increases the ratio of walks to non-walks without significantly affecting how well the hitter performs if he isn’t actually walked. In other words, if you’d rather have a walk than whatever the hitter normally does if he doesn’t walk, then you should walk him. If you’d rather have what he normally does, you should pitch to him normally.
|The Book Says: If a pitcher is trying to avoid pitching to a hitter, the hitter is significantly more likely to draw a walk, and moderately more likely to strike out. Specifically, a good, unprotected hitter in a good intentional walk situation is about 25% more likely to walk than the same hitter in a bad intentional walk situation, as well as about 10% more likely to strike out. Even an average hitter, with an average hitter on deck, is 20% more likely to draw the walk if the situation is a common one for intentional walks, and about 5% more likely to strike out. However, if the ball is hit into play, the pitcher’s approach (pitching to him, versus pitching around him) has no significant effect on the hitter’s statistics.|