Just because you have the 4th fewest sacrifice hits in MLB doesn’t mean you know what you are doing.

Posted: October 20, 2013 in Game Theory, In-game strategy, Managers

I’m talking about John Farrell and the Boston Red Sox. They had 24 sacrifice bunt attempts during the regular season, the 4th fewest in baseball. I don’t know how many they attempted or where they rank in attempts.

In game 6 of the ALCS, Boston attempted 2 sacrifice bunts, one with Victorino and runners on first and second, and one with Drew and a runner on second. With Victorino the game was tied, and with Drew the Sox were down by a run.

There is nothing necessarily wrong with both of those attempts. As I have always said, in a potential bunt situation, if the batter is good bunter and fast (I assume both of those batters are), he can bunt some (specific) percentage of the time on a random basis, as long as the infield is not overplaying one way or another. If the infield is playing optimally, according to game theory, then it doesn’t matter whether the batter bunts or not – the win expectancy (WE) should be the same for both strategies. That is the definition of the defense playing optimally – making the offense agnostic as far as bunting or hitting away is concerned.

Now, it is possible that even if the infield is playing up as far as they can, the bunt can still have a higher WE than hitting away. I suspect for that to be the case, the batter has to be a very poor hitter and an excellent bunter with good or great speed. It is also possible for the defense to be playing back all the way yet the WE for hitting away is still greater than the WE for bunting. That is often the case with good hitters at the plate who are also not good bunters and/or they are not fast. However, if the defense is playing anywhere but all the way back (as they would if it were not a potential bunt situation) or all the way in, the assumption is that they are playing in a configuration such that the batter can bunt or not bunt and the WE is exactly the same. If that isn’t true, then the batter must either bunt a lot (if the defense is playing too far back) or hit away a lot (if the defense is playing too far in).

Back to these two situations. The thing about the defense and the WE (of both bunting and not bunting) is that the latter is not static throughout the PA. As the count changes, so does the WE for both the bunt and hitting away, especially hitting away. That is obvious, right? If the count goes to 1-0, the batter becomes a better hitter. To a lesser extent, even if the defense remains the same, even the WE of the bunt attempt probably goes up. One, you are more likely to get a buntable pitch, two, if you bunt foul or take a strike, you are now 1-1 rather than 0-1, and three, since you don’t have to offer at every pitch even when bunting, you are more likely to ultimately draw a walk when attempting to bunt at a 1-0 count.

As the count changes, the defense should move to reflect the fact that the WE from hitting away likely changes more than the WE for bunting. If the count goes in the hitter’s favor, they should move back. It is not so much that they now anticipate the bunt less often, although they should, it is just that they want to play in such a way that they make the WE from the bunt and hitting away exactly the same – and that requires moving back in hitter’s counts (and up in pitcher’s counts other than with 2 strikes). So really, even when the count changes, the batter should still be agnostic as far as bunting or hitting away in concerned – it shouldn’t matter what they do.

But, we all know that managers often employ less than optimal strategies, especially when it comes to the sacrifice bunt, both on offense and on defense. It is likely that the defense did not move back when the count when to 1-0 on Victorino and 2-1 later on Drew. If they did move back at the 1-0 or 2-1 count, then either the bunt or the non-bunt would be justified. Let’s assume that the defense didn’t move though. And let’s use run expectancy (RE) rather than WE to for my analysis, just for simplicity sake.

In a low run environment, the RE with runners on first and second and 0 outs is around 1.5 runs. Let’s assume that that is the case with the defense playing a little up in anticipation of a possible bunt. If the defense is playing optimally, the RE for the bunt and hitting away should both be 1.5 runs, given the batter, pitcher, fielders, etc. Again, at that point it doesn’t matter whether the batter bunts or not. Now the count goes to 1-0. How much does that affect WE? At a 1-0 count, instead of a RE of 1.5 runs, it is around 1.56, so somehow the bunt has to be worth at least that much for it to be correct to bunt. The only way that is possible, assuming that the bunt and hitting away had the same RE when the AB started, was for the defense to back up at the 1-0 count. Even if the defense did move back, for the offense to be playing optimally according to game theory, when the count goes to 1-0, the batter has to hit away more often!

In case you are actually able to follow this, you might be asking, “Why must the offense still bunt and hit way in a certain proportion even when the defense makes them agnostic to their own strategy?” If they don’t, the defensive team can change their positioning at some point before the pitch arrives at the plate or the batter gives away his intention. As well, it tips off the defense the next time this situation comes up, although you can change your strategy to account for that.

The other thing is that Victorino used to be a switch hitter. In fact, I could swear that he hit from the left side in game 3 or game 4. If Victorino bats from the left side, the RE from hitting away with runners on first and second is higher for 2 reasons: One, fewer GDP, and two, he moves the runners over more often on an out.

Which brings up the second instance with Drew at the plate and a runner on second only. With a runner on second and no outs, the RE is around 1.13 runs. At a 2-1 count, it is 1.18. So, you have a similar situation as you had with Victorino. If the defense does not change their position with the count, you must switch to hitting away (at least a greater percentage of the time), and if they move back, you still must hit away more often, on a random basis. Again, I doubt that Detroit changed their defensive alignment. I am pretty sure that Jim Leyland was absent from class on the day that they went over game theory. And of course, before the count went to 2-1, it started out at 1-0 and then 1-1. The 1-0 count, as with Victorino, was another good time to switch to hitting away (you could then switch back to bunting at 1-1 and then not bunting again at 2-1, although with this kind of strategy you risk being too predictable).

The worst part about this bunt was that Drew is a lefty. I don’t know why lots of managers insist on bunting runners over from second base with a lefty batter. Surely they realize that he is going to move the runner over on an out when hitting away a significant percentage of the time. With a lefty batter and a runner on second, even if he is a good bunter and fast, you probably want to bunt much less often if at all. And the defense should play accordingly (not nearly as far in as with a comparable – in hitting and bunting ability, and speed – righty batter), in which case the offense would be agnostic as to their strategy.

To give you an idea of the difference between having a lefty and righty batter at the plate with a runner on second and no outs, here are the respective RE’s (there is no guarantee that they of equal hitting talent of course):

RHB: 1.104

LHB: 1.157

That is a pretty big difference. So, the RE from bunting if you are a left-handed hitter like Drew (and Victorino if he batted lefty) has to be a lot higher in order to justify a bunt attempt, as compared to a right-handed batter. Combine that with the 1-0 or 2-1 count and the bunt becomes questionable. Then again, it depends where the defense is playing, as always. If they are playing optimally, given the handedness of the batter (along with everything else), then it doesn’t matter what the batter does. And so the defense cannot take advantage of the offense, the batter must bunt and hit away in some exact proportion which makes the defense agnostic to their positioning (wherever they play, the RE from the bunt/hit away strategy of the offense is the same).

By the way, does that pitch from Veras go down in post-season history as one of the most predictable and worst location pitches on an 0-2 count ever? You probably have to throw the fastball more than you normally would at that count because you cannot afford to bounce a curve ball (especially with the gimpy Avila behind the plate) and you surely want to throw the curve ball in the dirt in that situation, if you choose to throw the curve ball.

  1. derek florko says:

    I just do not understand, these guys are supposed to be the best game managers in the world. It is so frustrating they have all of this info available to them, just look at the damn numbers how hard is it! instead of making up outside variables to justify the bunt why dont you do what has made your team successful all year long! I always tell my players and other coaches that I could manage a game better that these “professionals” and I constantly get laughed at, but I truly believe that.

  2. MGL says:

    Managers are not hired for their strategy acumen or even for their ability to recognize what it is they do not know about strategy. They are hired for their (supposed) ability to lead and motivate players and to make quick and decisive decisions even if those decisions are wrong!

    Also, the people who interview these managers typically don’t know anything about these kinds of strategies either.

    Finally, there is no potential manager that has any idea what to do in situations like this, nor do they have any idea how to even frame or analyze the questions.

    However, IMO, It is up to the people in the front office to train their managers to utilize the correct strategies by providing them with the correct information.and tools to do so. I don’t really blame the managers. Most people think that they know a lot more than they actually do. Managers are no exception. In fact, they are probably huge offenders! I blame the front offices. Some of them actually know about this stuff and have the ability to train their managers.They feel that it is not worth the effort because they feel like the other skills and qualities that make for a good manager are much more important and they don’t want to risk alienating or distracting the manager or the players, or they feel like it is not worth the time and financial investment.

  3. Florko says:

    I can totally agree with you, which brings up a new question, is the ability to “motivate” and “lead” more important than the in game strategies they try to employ? If you for instance were on the bench instead of John Farrell this season do you think the Sox would have won more games.

  4. MGL says:

    Totally unanswerable questions. I’ve guessed over the years that the difference between a “perfect” decision-maker and the average MLB manager is 2-3 wins per season. Wild guess. What is the spread between managers? I don’t know. 1-2 wins? Totally depends on what we are including. Lineups, personnel, sac bunt strategy, IBB, bullpen management, pinch hitters, defensive alignment, pitchouts, suicide squeezes, how long to let the starter go, hit and runs, SB attempts…

    If I am on the bench of course my win expectancy for the season would be greater than any manager’s. Keep in mind that I don’t know the answers to any of these strategy questions any more than any managers do (unless of course I have already researched them or someone else has, and even then, they will often be personnel and situationally dependent). The difference is that I have a better awareness of what it is that I don’t know, and I am unwilling to just make some stupid guess based upon tradition or faulty or ignorant reasoning. I know how to frame the questions and how to go about finding a solution and I am willing to do so before they come up in a game.That is what I would call, “Being prepared” at least in that realm. And if I owned or was the GM of a team, I would probably have a bench coach just for that although I would certainly educate the manager about such things.

    If I am the actual manager for the whole year? I have no training at that and I would have no credibility among the players. Of course that would likely be a disaster, but you never know. It is one of those hypotheticals that is impossible to answer. What if we made Bill James (a smart, sensible guy with lots of knowledge besides baseball) President of the U.S.?

    That would make for a good reality show, huh? MGL as manager of a team for a year? I did play and coach youth baseball for many years, so it’s not like I have absolutely no experience whatsoever!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s