The Rational Guide To Bunting (Curtis Granderson Edition)

Here’s how Girardi defended his decision: “I was trying to get a lead, in a sense … [Mets’ reliever Tim Byrdak] is a lefty [Granderson] hasn’t seen a lot of. I thought it was a good time to bunt there. There are times where I will just let him swing the bat there, as well. You’ve seen me do that. But today I thought it was time to bunt.”

Uggh. There may be a defense in Girardi’s explanation, but I can’t find one, so it falls to me to provide one. I think Girardi made the right call here. At minimum, it wasn’t the wrong call, it was a defensible call, and there are strong arguments in favor of having Granderson bunt.

Let’s start with a couple of rational rules. These rules are general rules, subject to exceptions, but they are (I think) good places to start.  Here is rule number one: bunting with a man on first and no one out is always a suspect move, always requires a robust defense, unless it’s the pitcher (or someone equivalent) who is doing the bunting. Again, let’s stress: a bunt in that situation can sometimes be justified, but it always requires justification, and in most cases the justification won’t be there. Why? Because a sacrifice bunt runs up against one of the sacred rules of sabermetrics: you don’t give away outs. Outs are precious. A team only gets 27 of them in a nine inning game; so long as the team has outs, it has a chance. Run out of outs, and it’s game over.

We look with great suspicion at the sacrifice bunt, because in general the out that’s given up is more precious than whatever we expect to gain if the bunt is successful. For example: when a team begins an inning with a runner on first and no one out, the team is expected to score about .9 runs that inning, and the team has a 43% chance of scoring at least one run in that inning. But if the team attempts a sacrifice bunt at that point and the sacrifice succeeds, the team’s run expectancy shrinks to .7 runs, and the team’s chance of scoring at least one run is reduced to 41%.

(Please note, I’m taking my numbers from Dan Levitt’s terrific article here at Baseball Analysts. You may find slightly different numbers reported elsewhere, but all the numbers I’ve seen on this subject are similar. As an example, I’ve checked the numbers I’m using here against those in “The Book: Playing the Percentages in Baseball” by Tango, Lichtman and Dolphin, and my numbers are close to theirs even if I don’t always reach the same conclusions that they do.)

But not all bunting situations are the same. Here’s rational bunting rule number two: bunting with a man on first and second and no one out is often a good move. Again, here are the numbers: with first and second, none out, a team’s run expectancy is about 1.5, and the team’s chances of scoring at least one run is 64%. With second and third, one out, a team’s run expectancy shrinks to 1.4, but the team’s chances of scoring at least one run increases to 69%. If the team needs one run (as arguably the Yankees did when Granderson bunted, with the score tied 3-3 in the bottom of the seventh), then the first and second none out sacrifice bunt looks like a good move.

We can drill down further than that, as Dan Levitt has given us splits we can use depending on who is bunting at different positions in the batting order.  However, these splits don’t change our calculations much – with the number two hitter in the lineup at bat, the first and second no out bunt still decreases run expectancy by a tiny bit and still increases the chance of scoring at least one run by about 5%.

Girardi’s decision to have Granderson bunt can be seen as a trade-off: a very small decrease in the chance for a big multi-run inning, in exchange for a small but substantial increase in the Yanks’ chances to score a single run. Does this trade-off make sense?  The answer depends on the game situation: if the game is close and the inning is late, the trade-off may make sense.  In “The Book: Playing the Percentages in Baseball”, authors Tango, Lichtman and Dolphin compared the win expectancy of different situations in (lucky for us) the bottom of the seventh with the home team one run down. With runners on first and second, no one out, the home team wins about 55% of the time … which is the same winning percentage as with a runner on second and third, one out.  So in this situation, again on average, the successful sacrifice bunt doesn’t help but it doesn’t hurt either.

Up until this point, we’ve discussed successful sacrifice bunts, where the batter is out and the runners advance.  But not all bunts produce this result: according to “The Book”, bunts with men on base result in a sacrifice about half the time. The bunter achieves a different but positive result about 19% of the time (anything from a walk to a bunt single to a two-strike swinging hit). Another 31% of results are negative: the batter strikes out, or bunts into a force play, or hits into a double play (either bunting or eventually swinging away).  But the positives here outweigh the negatives: the average run expectancy following a sacrifice attempt (ignoring all other factors, including number of outs and runners on base) is about .13 higher than that following a successful sacrifice where the batter is out and the runners advance.

The authors of “The Book” ask us to imagine a situation where we want our team’s hitter to execute a sacrifice bunt. Let’s say that the manager of the other team offers us a deal: to speed things up, we’ll declare our batter out and have the runners all move up a base. “The Book” tells us that this is a bad deal; we’re better off proceeding to let our batter bunt. Putting aside the possibility of the sacrifice, more good than bad is likely to flow from the bunt attempt; there’s a small advantage in trying the bunt and seeing what happens.

We need to keep this small advantage in mind when we think about the decision to bunt.  Three paragraphs ago, I noted that the win expectancy holds at about 55% when a batter lays down a successful sacrifice bunt with runners on first and second and none out in the seventh inning of a close game. But because other good things can result from a bunt attempt, there is a slight advantage in having a hitter try the bunt in that situation.

But we can do better than this. We don’t have to rely solely on comparing run expectations with men on different bases with no outs versus one out. We can turn back to the Dan Levitt piece, and look at the actual run expectancy following real-life bunt attempts.  What happens in the American League, when a team has runners on first and second, no one out, and the number two hitter in the order is at bat? Here’s our answer: in the situation faced by Granderson in yesterday’s 7th inning, (1) the hitting team can expect to score 1.6 runs on average, and at least one run 68% of the time; (2) if the batter successfully executes a sacrifice hit, the expectancy climbs to 1.7 runs, and the chance of scoring at least one run increases to 76%; and (3) if the batter bunts, the expectancy climbs a little higher, to about 1.75 runs, and the chance of scoring at least one run is 74%.

Why does bunting in the real world produce better results than we’d expected? There are two reasons: (1) as we noted, good things (other than the classic sacrifice) can come from bunting, and (2) managers seem to choose good spots to execute these bunts.

So on average, it looks like a good strategy to bunt with men on first and second and no one out, at least when an average number two hitter in the lineup is at bat.  Of course, the guy hitting second in the average lineup is not typically a guy like Granderson, the second leading home run hitter in baseball, with the best wOBA and wRC+ in his team’s starting lineup. But we have to think about Granderson’s chances yesterday in the bottom of the seventh, facing the Mets’ Tim Byrdak.  The Granderson versus Byrdak confrontation was lefty against lefty. In past years, Granderson has had trouble hitting lefties; this year, Granderson’s slash line (batting average/on base percentage/slugging percentage) against lefties is a healthy .255/.314/.809. But consider: Grandy’s hit 8 home runs against left-handed pitchers so far in 2011, and he has four other hits total against lefties!  That’s pretty much a record of all or nothing at all. Make of that what you will. More important is that Byrdak has been extremely effective this year against left-handed hitters, with a .231 batting average against, a 1.26 xFIP and a 14.85 K/9 ratio. Small sample size, but so far in 2011 Byrdak has struck out roughly half the lefties he’s faced.  Girardi had reasons not to bat Granderson against Byrdak.

Granderson may be a stronger hitter than the typical lefty faced by Byrdak, but it can be good strategy to have strong hitters bunt, particularly when the strong hitter is fast and a good bunter.  According to “The Book”, strong hitters make better bunters because the defense expects weak hitters to bunt and strong hitters to swing away.  When a good hitter is at the plate, on average a bunt attempt can produce nearly as high a run expectancy as swinging away, provided that the defense is not expecting a bunt.  I do NOT think the Mets were expecting Granderson to bunt yesterday!

We also have to consider that the Mets’ bullpen has been very effective this year, and that “The Book” says to bunt more often in a low run-scoring environment.

So, let’s add things up. On average, bunting with two on and none out is generally a good idea with the number two hitter up. Granderson was facing a tough lefty. The Yankees had reason to play for one run, given the tie score and the late inning and the strength of the Mets’ bullpen.  From this, I conclude that the Granderson bunt decision was defensible. It was also debatable.  Most intelligent and well-reasoned decisions to bunt are debatable decisions. The authors of “The Book” put it this way:

Unfortunately or fortunately, depending on your point of view, analyzing the efficacy of the sacrifice bunt in the various situations is so complex and difficult and the results are often so close, that we can offer only a few clear-cut rules of thumb and a myriad of recommendations built on somewhat shaky foundations.

So, debate away! I initially hated the decision to have Granderson bunt. I’ve grown to like it better. I like that defenses now have to consider whether Granderson might bunt, even in situations where one would expect a slugger like Granderson to swing away. I like Granderson’s chances of beating out that bunt, or of Mets’ third baseman Willie Harris throwing that bunt into the seats behind third base. I like a bunt when there’s speed at the plate and speed on the bases (Francisco Cervelli has good speed for a catcher). I don’t see how Cervelli could have been forced at third so long as the bunt was halfway decent and down the third base line; the Mets could not call for a “wheel play” with a hitter like Granderson at the plate.  If Granderson missed the bunt or took a ball after showing the bunt, Harris would have had to play in a few steps closer. I like the unexpected quality of the bunt, the fact that it might have put the Mets a bit off-balance.  I like that the decision neutralized the Mets’ best left handed reliever. I like the Yankees’ chances with men on second and third, one out, and the 3-4-5 hitters in the lineup coming up. I acknowledge that Teixeira might get walked in that situation, but I’m happy with A-Rod and Cano coming to the plate with bases loaded.

If you disagree … well, that’s what a comments section is for. But comment rationally.  Consider the numbers, and all of the circumstances.

Oh, and by the way: next time Girardi has Brett Gardner bunt, with a man on first and no one out, with the defense expecting bunt and with Cervelli on deck … you can draw on every argument above to rip the bunt decision to shreds.

23 thoughts on “The Rational Guide To Bunting (Curtis Granderson Edition)

  1. I'd just like to say I agree wholeheartedly. There were a lot of strong opponents on Twitter yesterday, but I think some of the reactions were a bit automatic in nature, since we've seen numerous stupid bunts already this year. I didn't want to say anything, since 140 characters is by far too little to describe all the relevant factors, as you've shown here. So thanks for this, it is a really good summary of all the nuances involved here. Also of note is that the WPA on the play was actually (slightly) positive at 0.3% (which climbs to 0.6% if you factor in that this caused the Mets to IBB Tex).

  2. Having granderson bunt was the right move because it eliminates the worst possible outcome in a tie game late. first and second, no one out, the worst thing that Granderson could have done was ground into a double play. leaving a runner at third with 2 outs. Granderson lays down the bunt, you have runners on second and third and one out with Tex, Arod and Cano coming to the plate. Even if they walk Tex (which they did) You have Arod coming up with the bases loaded, which is a situtation he thrives in.

  3. Apparently it's a situation he thrives in even when he doesn't as evidenced by his nubber yesterday. I'll take it.

  4. Fw little IW, I thought it was a great call and a good bunt. Granderson was even close to running it out; the way balls were being handled, one could almost make the case that he was bunting for a single. Even if not, he laid down a great bunt that was unexpected, that put runners on 2nd and 3rd wtihout a double play. (or, a popup foul that is caught, still yielding an out, without advancing the runner.)

    Grandy batting .255 against lefties? – from my HS days, that means a 1 in 4 chance of a hit; odds are better for him to execute a successful, surprise, sac bunt. Good call, Joe.

    (and just btw, aren't some of the MSM folks complaining about the bunt ALSO the ones complaining about too many home runs?)

  5. Quick note. If you think of the tradeoff as 1st and 2nd 0 out versus 1st, 2nd, 3rd 1 out (as was the case after the IBB), the expected runs scored is higher in the second condition.

  6. I thought it was a bad call yesterday and while #'s are intriguing I still disagree. The numbers don't consider the likelihood that Granderson and Arod would've GDP. Granderson has ground ball rate this yr of 31% and Arod 45%. And let's face if Arod hit that ball a little harder that's exactly what he wouldve done. I cant condone a play that effectively removes that bat from arguably our two best hitters and doesn't even remove the rally killing DP from the equation.

  7. I love my readers! Didn't even think to check if Granderson had faced Byrdak before. Yes, this is a tiny sample size of at bats, but the information is worth a little somethng — it's a very small push in the direction of a bunt. Thanks for adding to the discussion!

  8. Another great point, thanks William. The charts I've relied upon are based on average run production in years where offensive production is higher than this year's. The more you read about "the year of the pitcher", the more advantageous it is to use the sacrifice bunt. I think this is a small factor — I don't think we've swung so far that I like the sacrifice bunt with a man on first and no one out. But in a close game against a team with a strong pitching staff, consider this to be another little push in favor of the sacrifice bunt.

    You also mentioned what I think is one of the more difficult points in this analysis: the tradeoff between playing for one run and shooting for the big inning. After the 15 inning marathon last week in Baltimore, we all have a higher appreciation for entering the final innings with a lead greater than one run. But as it turns out, the Granderson bunt decision (in my analysis) increased the Yankees' chances of scoring one run, and probably gave a small boost to the team's chances of scoring two runs as well. This is a difficult decision to make, as I keep saying. At least it should be a difficult decision. When the decision is an obvious one, that's usually because it's obvious that you should not bunt!

  9. Larry after rationally sitting down and thinking about the bunt then reading your article. I 75% like the bunt. On Friday their set up man and closer shut us down and we had no chance of scoring one run. And with the game tied and grandy coming up. You are right to say who would expect a bunt. The bunt could have gone wrong but it didn't. So I'm okay with the bunt as long as it leads to scoring. But I won't be okay making grandy bunt with runner on first and no outs. Giradi has done that before this year. I don't recall the specific game but he has, I understand the bunt now. But just because it worked out, I'm hoping Giradi doesn't stick with being mr bunt. This time I'm ok without the bunt due to how late it was in the game but next time I don't know if I'll be ok with the bunting. As for gardner, he needs to work on his bunting!

  10. Interesting. I truly enjoyed the article, it made a lot of great points. What was even more enjoyable was the lively discussion that ensued. Very well done by all that participated! Thank you, I wish more of these comment sections were occupied by such thoughtful, well reasoned and civil fans of the game. Btw, the bunt was a good move and you can't really expect a manager to explain the thought process he uses for game strategy to the media, can you?

  11. In Bill James book on managers, he makes an argument that the numbers on bunting by position int he batting order aren't very convincing. Can't remember why right now. In any case, Granderson is a much better hitter than the average second-spot hitter, so I think you're giving up more than you're taking into account.

    Playing for one run in the 7th inning seems like a bad move. The value of a run in the seventh inning of a tie game is much less than that of a run in the ninth of a tie game.