THE BOOK--Playing The Percentages In Baseball

The only reason why this could be a good move is if the run variance from the runner on 3rd-1out situation is less than that of 2nd-0 outs. 

What do the bell curves for runs scored look like in each situation?  Put another way, what are the odds that they score 0 runs from either scenario?  Is this a risk averse play?

Why do you care about run variance?

***

Chances of a scoreless inning:
2b, 0 outs: 37%
3b, 1 out: 34%

So, you have a bigger chance of scoring with a runner on 3B and 1 out, than with 2b and 0 outs.  But so what?  This is the first inning, not the 9th.

We want to count all the runs we can:
2b, 0 outs: 1.19 runs to end of inning
3b, 1 out: 0.983

Difference = .207 runs

Since that’s roughly equal to .021 wins, this is in-line with the win probability numbers.

***

What we care about is win probability, and not run expectancy and not scoring probability.

I was wondering if it was the correct decision if you wanted to be risk averse.  Just curiosity.

Right.  I mean its the right move if its late and say its a tie game and your playing to score exactly 1 run.  But in the first inning, I hate it.

I know its not the exact situation but the mets do it all the time first inning.  Reyes singles, Luis bunts him to second.  It drives me insane.

There is no discussion unless one considers the distribution of possible events from a bunt attempt AND one considers the game theory aspect, which Tango talks about.  Of course, as Tango also mentions, for there to be any game theory advantage, Barton has to sometimes bunt and sometimes not (probably not bunt a lot more than bunt) such that the corners are forced to play closer than they would if he never bunted (so that when he does NOT bunt, he gets more hits).

Without a discussion (and quantification) of those two things (and other things), there is NO discussion. None.  Zero.

Telling is the WE of a runner on 3rd with 1 out out versus a runner on second with 0 outs tells us nothing.  Nothing.  Zero.  Nada.

If Poz, a self-admitted sabermetrically-leaning sports-writer, would just read The Book (if he has not already) and if he would refer to it when he has an issue like this, he could discuss that issue intelligently.

In general, it is probably NEVER correct for any player to sac bunt unless he is a good bunter with good speed, otherwise he will not get enough singles and ROE’s to push the WE from a bunt ATTEMPT past the WE for hitting away.  Of course, as everyone knows intuitively, the earlier in the game, the less you tend to bunt, but it still depends on the distribution of results from a bunt attempt.

A “successful” (runners advance, batter makes out) is NOT a good proxy for a bunt attempt.  Please write that on the blackboard 100 times!

A slow runner and/or bad bunter, and pitchers, do substantially worse overall than a “successful” bunt.  A good bunter and/or fast runner does substantially better than an out and a runner advance.  So how can you use the assumption that the batter makes out and the runners advance to analyze a specific bunt attempt? You can’t!

And that is not even considering the game theory aspect, which has already been explained to some extent above (basically if the decision is close enough, when you sometimes bunt and sometimes don’t, you force the infield to have to play up a little, this increasing your WE when you hit away).

So, in summary, in order to come close to determining or even having an “opinion” on whether a batter should bunt or not, you need to estimate the following:

1) The WE from hitting away, which should be fairly easy to do.  And even then, if it ends up that the decision whether to bunt or not is close, the WE hitting away goes up a little when you bunt some of the time, thus increasing the overall WE from a “sometimes bunt” as opposed to a never bunt.

2) The approximate distribution of results of a bunt attempt, which are, out and runner advances, gets to 2 strikes and has to hit away, walks, singles, ROE, bunts into an out, no runners advance, or bunts into a DP.  Then the WE for each of those results needs to be weighted and averaged, for the overall WE of a bunt ATTEMPT.

Then and only then can you have an “opinion” on the bunt attempt.  Again, if it is close, you can tweak the numbers.

From the many, many hours of research I did for the bunting chapter in The Book, I can tell you that for good bunters who are fast, the WE’s for hitting away and for attempting a bunt are going to be close at any point in the game.  For bad bunters and/or slow runners, they are usually not that close.

As it turns out, the hitting ability of the batter does not matter that much.  That is only because for a poor hitter, the infield expects a bunt such that the bunt attempt does not include a lot of singles and ROE’s and more runners getting thrown out at second or third.  With good hitters, the infield tends to play back so that when they do bunt (assuming they are good bunters and are fast, like an Ichiro), they get a lot of singles and ROE’s and the lead runners don’t get thrown out as often.

If you want “rules of thumb” go to that chapter in The Book, and you will find them.

I can, however, tell you off the top of my head that unless Barton is a very good bunter and has good speed, he should never bunt, especially as a left handed hitter with a runner on second (or first and second), since he will obviously move the runners over more often with a ground ball than will a RHB, and he will hit into fewer GDP than a similar-in-speed RHB.

Telling is the WE of a runner on 3rd with 1 out out versus a runner on second with 0 outs tells us nothing.  Nothing.  Zero.  Nada.

I disagree.

It tells us whether or not the team/batter had the right idea.  If we are operating under the assumption that the player is intending to make an out, and not trying to get on base via a “surprise” bunt, then the WE tells us whether the intention was a correct strategy. 

In other words, if a team could theoretically trade a base for an out in that situation:

a) Would the a’s do it?
b) Is it correct to do it?

That’s what it tells us.  Irrespective of results, it is interesting to analyze what teams are trying to do, and whether or not they have the right idea.  If they are trying to sacrifice, and manage to ROE, they are merely pulling a Homer.

Well, trading an out for a base ALWAYS lowers your RE and almost always lowers your WE, so that is not very useful or interesting information, is it?

MGL:  I’m curious, has there ever been a player in the history of baseball who is a good enough bunter to make it correct to bunt with no outs and a runner on 2B in the first inning, who is also a LHH with average or worse speed?  If not, in this case it seems like we can safely ignore all the things you insist we must not ignore.

The problem I see with the game theory analysis is that bunting when it isn’t optimal should never make the defense play closer than normal. If when playing back, the defense would rather have Barton bunt than hit away (which is true given WE), then there is little to be gained by playing in.

I’d be interested to see some evidence that slow players bunting more often actually affects where the defense plays.

"I’m curious, has there ever been a player in the history of baseball who is a good enough bunter to make it correct to bunt with no outs and a runner on 2B in the first inning, who is also a LHH with average or worse speed?  If not, in this case it seems like we can safely ignore all the things you insist we must not ignore.”

Yes, of course.  If the defense is not expecting a bunt.  If the defense is playing back, almost anyone who is a decent bunter can bunt such that the WE is higher than hitting away.

In any case, I did say this:

“I can, however, tell you off the top of my head that unless Barton is a very good bunter and has good speed, he should never bunt, especially as a left handed hitter with a runner on second (or first and second), since he will obviously move the runners over more often with a ground ball than will a RHB, and he will hit into fewer GDP than a similar-in-speed RHB.”

Ken, if Barton were to bunt 100% of the time, even if it is incorrect, where do you think the proper defense alignment is?  All the way in of course.  If the offense is acting optimally, then it doesn’t matter where the defense plays. If the offense is bunting too much, then the correct play is to play in somewhat at least.

Didn’t Ortiz get the shift on, and he countered that by bunting.  He was like 3 for 4 or something on the bunt.  Did I just make that up?

But Barton has not reached a single time.  He is sacrificing, not trying to get a surprise bunt hit.  I haven’t watched the video, but I bet he squares early, giving him almost no upside here.  (How does official scoring on a sacrifice work—if the hitter is attempting to bunt for a hit but is out, and runner advances, does it get scored as a sac hit?)

"Didn’t Ortiz get the shift on, and he countered that by bunting.  He was like 3 for 4 or something on the bunt.  Did I just make that up? “

He’s attempted to bunt (not always putting the ball in play before reverting) a few times this season.  I don’t remember him ever trying in years past.  He hasn’t made the defenses change their strategy as far as I can tell.

“(How does official scoring on a sacrifice work—if the hitter is attempting to bunt for a hit but is out, and runner advances, does it get scored as a sac hit?) “

If you advance a runner it’s always a sacrifice.

In practice, a SH might almost always be credited if runners advance, but the scorer is granted a measure of discretion:

10.09(d)Do not score a sacrifice bunt when, in the judgment of the scorer, the batter is bunting primarily for a base hit and not for the purpose of advancing a runner or runners.  Charge the batter with a time at bat.  NOTE: In applying the above rule, always give the batter the benefit of the doubt.

When batters are trying to sacrifice, they get hits or ROE around 8-10% of the time. If you are a good bunter and/or fast, you get more.  If you are a bad bunter and/or slow, you get less. If the defense is highly expecting a bunt, you get less.  If they are not, you get more.

It is the same with the other possible results of a bunt attempt. If you are a good bunter, you hit into fewer force outs, and you get fewer 2 strikes where you generally switch to hitting away (and have a bad PA since there are 2 strikes).  The result of a bunt attempt even depends on how good of an eye you have.  The better your eye, the more you will walk when attempting a bunt.

Again, all of these things have to be considered when analyzing whether a bunt attempt is correct or not.  And yes, it is true (obviously) that we can sometimes tell when someone should definitely NOT be bunting without looking at all of these things.  Basically if you are not a good bunter and not a horrendous hitter, you should never be bunting.  And of course, where you draw the line (how bad a bunter and/or how good a hitter you have to be not to ever bunt) is different early and late in the game.

I don’t know that it is all that obvious without looking at the numbers, that Barton should not be bunting in the first inning with a runner on second.  I would say off the top of my head that he would have to be an excellent bunter with decent speed (maybe a 12% chance of getting a hit or an ROE on a bunt) in order for him to bunt at least some of the time.

Also remember that if it is at all close, even if the bunt is worse than the non-bunt, it still may be correct to bunt a small percentage of time in order to force the IF up a little (now the non-bunt is worth more by virtue of you bunting a little).  Maybe.  Again, you have to “run” the numbers.

I would also say off the top of my head that it is probably correct for any decent bunter to sometimes bunt in any possible bunt situation at any point in the game.  I say that because if the infield is playing completely back (as if the batter is never going to bunt), the WE from a bunt attempt is almost always higher than the WE from not bunting, even for a good hitter.

mgl, I’m not sure that batters that are attempting a SH maintain the threat of a swing long enough to enjoy any meta benefits when they are swinging.  What do you think?

It seems like there’s a fundamental difference between bunting for a hit (where you maintain the threat of a swing until very late but get a poorer quality bunt) and a sacrifice.  In sacrifice situations, maybe batters should be bunting for a hit at a much higher frequency but rarely sacrificing.

Maybe I don’t really understand the situation, but I think even good hitters should be bunting for a hit on <5% of pitches in SH situations, and maybe mediocre hitters should be bunting for a hit on 40-70% of pitches

"mgl, I’m not sure that batters that are attempting a SH maintain the threat of a swing long enough to enjoy any meta benefits when they are swinging.  What do you think?”

I don’t know what you mean.

For what it is worth, the reason that batters generally do not try and bunt for a hit in sacrifice situations is this:

The difference between bunting for a hit and and a sacrifice is two-fold.  One, when trying for the hit, you square as late as possible in order to keep the infielders from charging in until the last possible moment.  Two, when trying for a hit, you are more particular about where you place the bunt.  When the infield is already expecting the bunt, number one is negated - there is no point in waiting before you square.  Even number two is somewhat negated when when the infield is already playing in or at least mostly in.

The only time you see a batter bunting for a hit in a sacrifice situation is generally when the infield is not expecting a bunt, as with a batter who does not usually bunt.

IOW, when the IF is expecting a bunt, the advantage you get from attempting the hit rather than the sacrifice is less than the disadvantage of trying to bunt for a hit (which are more foul balls and more missed bunts).

Tango/11: I don’t know about Ortiz, but a few times as a Yankee Giambi - when struggling in 2005 - would attempt the bunt.  FanGraphs has 3 attempts and 2 hits, but that doesn’t speak to times he bunt foul, missed, etc.  I don’t think the defense ever changed strategy though, and also pretty sure he always did it with none on.

Hinske and McCann have bunted for hits against the shift this season, and it has affected where the defense plays them, at least in my eyes; I think McCann is seeing fewer shifts than last year.

McCann has also got himself into a few 2-strike holes by bunting foul.
---
If a defense knows a guy likes to sacrifice bunt but never tries to bunt for a hit (like Barton, apparently), will the threat of bunt affect where the defense plays him?  I think JB H/16’s point is that the defense would not alter their positioning, knowing it should still be easy to get him out in a sacrifice when he shows early.

I disagree; the infield would still be likely to play up a bit in hopes of getting the lead runner or more easily geting to a bunt in the air.

If defensive positioning were determined by computers with accurate hit charts (split along pitch type, pitcher handedness, count, etc.) then random bunts would probably be optimal (for some/most players) from the game theory perspective.

But I think with the way that defensive positioning is actually determined (some combination of primitive metrics, received wisdom, and managerial whimsy), it’s too close to tell if you are gaining a marginal long-term edge or giving away outs.

A simpler, somewhat comparable thing is pitch outs. I don’t remember if a pitch out (at the observed rate) is on average worth giving up a ball, but if the other team knows you will never pitch out they will likely steal more often.

If a batter can either

A) Square early and sacrifice
B) Swing away
C) Bunt for a hit

Then the frequency of A) shouldn’t have any effect on what the defense does when the batter chooses B) or C), right?

#21, don’t know what you mean.  Where the defense plays depends on the percentage of times that the batter does A, B, C, etc. (usually only A or B) in that particular situation.

Don’t forget this:

If the defense plays optimally against an optimal opponent, then it will play in such a manner that the WE for all possible alternatives by the batter (sac, hit away, bunt for a hit) yield the same WE.  If the batter is not playing optimally, then the defense must choose how much to take advantage of the non-optimal play by the batter.  The reason they choose is this:

Let’s say that the batter bunts just a little less than he should. It is then correct for the defense to play all the way back.  But if they do that, surely it will encourage the batter the bunt more, which would be more optimal for them.  So, on defense, if the batter is bunting too little, then you still play a little up in order not to encourage him to bunt more which would push him in the optimal direction.

It is like you play against the amateur poker player who bluffs too much. You are supposed to call a potential bluff against him 100% of the time, but you can’t do that or he will quickly stop bluffing or at least as much as he does.

"#21, don’t know what you mean.  Where the defense plays depends on the percentage of times that the batter does A, B, C, etc. (usually only A or B) in that particular situation. “

My point is that the batter announces their intention to either do A or (B + rarely C) well before the ball is put in play

Isn’t the defense either fully prepared to defend a sacrifice or fully prepared to defend a swing by the time contact is made?

"Isn’t the defense either fully prepared to defend a sacrifice or fully prepared to defend a swing by the time contact is made?”

No.

Guy/12 - Barton is sacrificing like the worst of the worst pitchers. Last night in the first inning against the Cardinals, he squared away. TWICE. He turns his body, gets in bunting position, and bunts.

Right now I’m considering nominating Daric Barton (who I otherwise like) as the dumbest player in sports. I actually wish the A’s would just demote him to Kane County for the rest of the year so he stops being a moron, but, not surprisingly, Bob Geren (who makes Dusty Baker look like Stephen Hawking) is encouraging this practice.

And Billy Beane is too busy watching the World Cup to notice or care.

I think that many of the sabermetric teams have a “hands-off” policy when it comes to the manager’s in-game decisions, at least during the season.

What I would like to know is how often Barton actually attempts to bunt in a potential sac situation, which would be 0 outs and a runner on first, second, or first and second.  Also, how good of a bunter is he?