Farrell and the Bunting Strategy

[nattysez]

OH MY GOD THE BUNTING.
 
When I've complained in the past about this team not having a philosophy, this is the kind of thing I meant.  When Theo left, Henry claimed that Bill James was going to start having more influence over the organization.  I doubt very much Bill James is advising them to bunt constantly.  
 
And if you're going to have JBJ bunt two guys over with the top of the lineup coming up, I can at least see some semblance of logic there.  But having the $10mm SS trying to bunt Gomes over ahead of Nava (who couldn't crack the lineup against a righty two days ago) and JBJ is simply indefensible.
 
Between the circus on the basepaths earlier this year (they are still only 18-for-29 in SBs), the Nava-Gomes bizarreness of the past week, and all of this bunting, I am really starting to wonder about Farrell.   

 

[Auger34]

I think the bunting speaks to the complete lack of trust Farrell has in the offense. Might as well play for one run or have a chance at getting a run with a fly out then fail at trying to string together a big inning

 

[nattysez]

I think the bunting speaks to the complete lack of trust Farrell has in the offense. Might as well play for one run or have a chance at getting a run with a fly out then fail at trying to string together a big inning

 
Bunting ahead of Nava and JBJ is not playing for one run - it's playing for none.  And bunting in general is giving away outs, which is the LAST thing you should do with a struggling offense. It seems like Farrell's decided to ignore years of research about the effects of bunting on run expectancy because the Sox have hit into a lot of double plays for two months, which is really troubling and ultimately going to accomplish the opposite of what he thinks it will.

 

[smastroyin]

I think it is more likely that Farrell believes that by executing fundamental small ball type plays the hitters will be helped out of their slumps. I'm not sure I agree with that, but I assume it is part of the equation.

 

[BosRedSox5]

We've known for years that a sac bunt in that situation is a bad play. We have the data. I thought it was well established fact that having a runner on second with one out gave you a lesser chance to score than a runner on first with no outs. It's totally indefensible to bunt in that situation, even if it had "worked" because in that case we'd he squandering a precious out for 90 feet of real estate.

Bunting a guy to third may make some sense, a squeeze play is the right call in some situations... but bunting a runner from first to second is and always has been a boneheaded move. 

 

[Savin Hillbilly]

It was especially weird because Carrasco is a righty with a normal platoon split. Asking Drew to bunt if a lefty had been brought in would be a bit more defensible, assuming you didn't want to PH for him.
 
I wasn't watching the game--was it pretty clear that the bunt was called from the bench and not Drew improvising?

 

[Smiling Joe Hesketh]

I think it is more likely that Farrell believes that by executing fundamental small ball type plays the hitters will be helped out of their slumps. I'm not sure I agree with that, but I assume it is part of the equation.

 
And there are a great many managers who believe the same thing. The problem is that it's wrong, it actively hurts the team when trying to score runs and he's doing it far too often to be an acceptable course of strategy.
 
This is what I mean about Farrell being a smart guy who makes baffling in-game decisions. Bunting in general is a stupid strategy that results in fewer runs than would be expected if he just let the batters swing away. Yet he continues to do it. With the team fading out of the playoff picture (8 1/2 games out this morning, and it's now June so it's officially no longer "early") he cannot be wasting outs and games pursuing less-optimal strategies like this.
 
Using Mujica last night, well I can't hold that against him. It was the 12th inning, he had already used his better bullpen options.

 

[smastroyin]

The only argument I am offering is that it must be a strategic, not tactical, decision.  Because there is no defense of it as a tactical decision and I would like to think that Farrell knows enough about the numbers to know that. 

 

[absintheofmalaise]

Broke this out of the SGC thread.

 

[lexrageorge]

The expectancy of scoring 1 run with man on 1st with none out is 44.1%.  With 1 out and man on 2nd it's 41.8%.  While lower, it's not dramatically lower, and there is some justification for playing for the single run there.  He couldn't run for Gomes, and Drew has only been facing MLB pitching for a couple of days, so it's not like he's taking the bat out of Bogaert's hands.  
 
Just saying the expectancy differences are close enough that the decision is not as horrible as it's being made out to be.  Of course, it backfired, but Nava and Bradley didn't exactly set the world on fire with their at bats either.  For whatever reason, this season's Nava is not last year's Nava, and I doubt Farrell has anything to do with that.

 

[pokey_reese]

But when the RE is lower, you can't cite "just playing for one run" as an excuse, because you are still reducing the chance that you score even that one run.  If there were something in the data that said "when you execute that sacrifice bunt, your odds of scoring one run increase but the odds of scoring 2+ runs decrease" but I haven't seen that data.

 

[ivanvamp]

Edit:  Forget it.  My question got answered while I was typing.

 

[lexrageorge]

But when the RE is lower, you can't cite "just playing for one run" as an excuse, because you are still reducing the chance that you score even that one run.  If there were something in the data that said "when you execute that sacrifice bunt, your odds of scoring one run increase but the odds of scoring 2+ runs decrease" but I haven't seen that data.

See the following:
 
 http://www.tangotiger.net/re24.html
 
Not everyone agrees with their methodology, but it's used by many folks here and elsewhere.  For example, when going from 1st-and-2nd with none out to 2nd-and-3rd with 1 out, the overall run expectancy goes from 1.556 to 1.447.  However, the chances of getting one run across increase from 0.643 to 0.698.  It is possible to play for 1 run as opposed to maximizing total runs scored in an inning.  And sometimes playing for 1 run makes sense (think 2004 ALCS).

 

[Bone Chips]

And those run expectancies are league average. When you factor in that the Red Sox seem to do a lousy job of executing bunts, the numbers get even worse.

 

[Reverend]

The only argument I am offering is that it must be a strategic, not tactical, decision.  Because there is no defense of it as a tactical decision and I would like to think that Farrell knows enough about the numbers to know that. 

 
I think you're right that it's strategic because, well, like you imply, it has to be. It's a strange strategic choice for a roster constructed to take advantage of OBP over AVG though; if one of the next guys walk, all you've done is wasted an out.
 
I've said this elsewhere, but I think he's pressing and trying to steal a win for the club. The dirty little secret about the seven game win streak is that the lineup didn't really start playing better overall; AJP and Bogaerts carried them through it while the Braves and Rays booted the ball around the infield. BROCK HOLT's bat has been a welcome addition, but the rest of the team was still not hitting well even during the streak and has not been (FG--then click "Last 14 Days" under "Split"), and they are not constructed to win games by leaning on two rookies and an aging catcher.
 
So I think he's trying to steal a game here and there for his guys. As pokey_reese points out though, if you keep trying that, the odd are you will blow more games than you steal, so while it's an understandable human attitude, it's still mistaken.

 

[Savin Hillbilly]

But when the RE is lower, you can't cite "just playing for one run" as an excuse, because you are still reducing the chance that you score even that one run.  If there were something in the data that said "when you execute that sacrifice bunt, your odds of scoring one run increase but the odds of scoring 2+ runs decrease" but I haven't seen that data.

 
Also, playing for one run in extras is not quite as cut-and-dried a strategy on the road as it is at home. For the home team, the difference between one and two runs is meaningless; not so for the visitors. So you'd want the bunt to produce a significantly higher likelihood of scoring one run, to outweigh the lesser overall RE.
 
If this wasn't a recurring theme, I would guess that Farrell had Drew bunt because he thinks Drew's timing is not back yet and he's just not very likely to catch up to 95+ heat, even from a RHP.

 

[Reverend]

See the following:
 
 http://www.tangotiger.net/re24.html
 
Not everyone agrees with their methodology, but it's used by many folks here and elsewhere.  For example, when going from 1st-and-2nd with none out to 2nd-and-3rd with 1 out, the overall run expectancy goes from 1.556 to 1.447.  However, the chances of getting one run across increase from 0.643 to 0.698.  It is possible to play for 1 run as opposed to maximizing total runs scored in an inning.  And sometimes playing for 1 run makes sense (think 2004 ALCS).

 
Turned off the game before it was over last night, did you?

 

[pokey_reese]

See the following:
 
 http://www.tangotiger.net/re24.html
 
Not everyone agrees with their methodology, but it's used by many folks here and elsewhere.  For example, when going from 1st-and-2nd with none out to 2nd-and-3rd with 1 out, the overall run expectancy goes from 1.556 to 1.447.  However, the chances of getting one run across increase from 0.643 to 0.698.  It is possible to play for 1 run as opposed to maximizing total runs scored in an inning.  And sometimes playing for 1 run makes sense (think 2004 ALCS).

Yes, and I was unclear in my previous post.  I was responding to your examples using a man on first.  Getting a man to third with fewer than two outs is the one situation where bunting might be defensible, but the problem is that Farrell often seems to be doing it to get men into scoring position, rather than when they are already in scoring position.

 

[twothousandone]

I think it is more likely that Farrell believes that by executing fundamental small ball type plays the hitters will be helped out of their slumps. I'm not sure I agree with that, but I assume it is part of the equation.

 

 
And there are a great many managers who believe the same thing. The problem is that it's wrong, it actively hurts the team when trying to score runs and he's doing it far too often to be an acceptable course of strategy.

It's wrong because it doesn't help hitters get out of slumps in the ensuing days, or because it hurts the team in that particular game?
 
 

If this wasn't a recurring theme, I would guess that Farrell had Drew bunt because he thinks Drew's timing is not back yet and he's just not very likely to catch up to 95+ heat, even from a RHP.

 
Is there an argument that having the batter bunt makes him pay more attention to watching the ball, everything about it, and doing that can be quickly rewarded with contact? Thus, the next at-bat or the next day, he's had a good habit reinforced?  I know I often read a batter comes out of a slump and says he's seeing the ball better. Is calling for a bunt a way to speed up the process of seeing the ball better?

 

[OttoC]

If there is a runner on 1st with no one out and the next batter fails to advance that runner, the team is left with a runner on 1st with one out (0.284), which is worse than a runner on 2nd with one out (0.418). So if you are looking to score just one run, what is the probability that the second batter advances the runner without sacrificing?

 

[MyDaughterLovesTomGordon]

I also have the impression Farrell is bunting like crazy, but the stats say the Sox have just 7 sac bunts, tied with the White Sox for fourth from the bottom. Texas has 22. League Average is 11. 
 
Maybe the worst part of Farrell's bunting strategy is that is consistently not working. Not only is he calling for the bunt too often, but it's FAILING too often, either through the player not getting the bunt down at all and having to swing away with two strikes, or someone getting thrown out at second (looking at you, AJ).
 
I'm not sure stats exist for failed bunt attempts, but I'd like to see where the Sox rank there. As with the stolen base failures, it seems like the Sox are not only employing bad strategy, but they're amplifying the problem with bad execution.

 

[joe dokes]

 
 
I'm not sure stats exist for failed bunt attempts, but I'd like to see where the Sox rank there. As with the stolen base failures, it seems like the Sox are not only employing bad strategy, but they're amplifying the problem with bad execution.

 
I haven't been able to find it either. And I have been shouting from the rooftops for attempts to be part of the regular stat package. It's one of the few things that involves both success and failure in baseball where only the success is tabulated, rather thn a rate.

 

[Savin Hillbilly]

I also have the impression Farrell is bunting like crazy, but the stats say the Sox have just 7 sac bunts, tied with the White Sox for fourth from the bottom. Texas has 22. League Average is 11.

 
7 successful sacrifice bunts, that is. That's out of 16 attempts, which is about league average. Our success rate of 44% is tied for 13th in the AL.
 
The numbers are here.

 

[Hee Sox Choi]

Using Mujica last night, well I can't hold that against him. It was the 12th inning, he had already used his better bullpen options.

 
You mean the SIX lefties they had in a row (I'm including AssCab switchy) that Mujica was just beginning to pitch to?  Bourn singled and A-Cab hit the jack.  There were 4 more lefties in a row coming up.  
 
However, I can understand why Farrell was avoiding Miller with his recent walk-off losses (and he threw 16 pitches the night before after 2 days off.
 
But this thread is about the HORRORS OF BUNTING and Farrell is our Colonel Kurtz leading us up the river of sacrifice.  

 

[smastroyin]

How do you deal with a failed (foul) attempt that then leads to a batter swinging away with 2 strikes?  Do you only want to count failed sacrifices as bunts that are in play but where the lead runner is thrown out?

 

[Reverend]

 
7 successful sacrifice bunts, that is. That's out of 16 attempts, which is about league average. Our success rate of 44% is tied for 13th in the AL.
 
The numbers are here.

 
Kansas City has fourteen bunt hits??
 
Edit: And the Mets are 26 for 27 on the sacrifice. This is offering some interesting perspective...

 

[Hee Sox Choi]

 
7 successful sacrifice bunts, that is. That's out of 16 attempts, which is about league average. Our success rate of 44% is tied for 13th in the AL.
 
The numbers are here.

 
But how many times did we fail on sac bunts to get 2 strikes on our guys and then strikeout?  That wouldn't show up in the #s.  So we've given up 16 outs and only moved the guy over 7 times?  Giving up an out to move a guy over is stupid, giving up an out to NOT move a guy over is infuriating.  Even if it is successful, the run probability drops but that doesn't take into consideration all the times it DOESN'T WORK (which is 60% of the time for us). So, statistically, it's an even more ignorant call to keep doing it.  Ugh.  

 

[joe dokes]

Thanks for the link. Dont know how I missed it for so long.
 
 

How do you deal with a failed (foul) attempt that then leads to a batter swinging away with 2 strikes?  Do you only want to count failed sacrifices as bunts that are in play but where the lead runner is thrown out?

 
Bref doesn't count those.
 
(only includes unsuccessful bunts made and bunt strikeouts)

 

[Savin Hillbilly]

 
You mean the SIX lefties they had in a row (I'm including AssCab switchy) that Mujica was just beginning to pitch to?  Bourn singled and A-Cab hit the jack.  There were 4 more lefties in a row coming up. 

 
You're leaving out the two righties (Raburn and Aviles) who started the inning. And Bourn is a noodle bat who historically has very moderate platoon splits (though they've been big so far this year). And you can count AssCab in either column. So really there were four guys that a RHP should have been able to get out before the real parade of lefties began. If I'm Farrell, I give the ball to Mujica and keep Miller in reserve for the 13th.

 

[Savin Hillbilly]

Bref doesn't count those.
 
(only includes unsuccessful bunts made and bunt strikeouts)

 
But it's true that because of this, the official tab of unsuccessful sacrifices is really only the iceberg tip of the strategy's failure to deliver.

 

[Hee Sox Choi]

 
But it's true that because of this, the official tab of unsuccessful sacrifices is really only the iceberg tip of the strategy's failure to deliver.

So it's even worse than imaginable.  I'm guessing a team with Bill James on it would know this.  This has to be all Farrell's evil doings.  

 

[MyDaughterLovesTomGordon]

 
But it's true that because of this, the official tab of unsuccessful sacrifices is really only the iceberg tip of the strategy's failure to deliver.

 
Thanks for turning me on to that stats page. Though it's depressing to see the Sox below average on getting the runner in from third and moving the runner over from second as well. Basically, the message is that - on offense - they do NOT do the little things well. Which I think all of us knew already.

 

[seageral]

See the following:
 
 http://www.tangotiger.net/re24.html
 
Not everyone agrees with their methodology, but it's used by many folks here and elsewhere.  For example, when going from 1st-and-2nd with none out to 2nd-and-3rd with 1 out, the overall run expectancy goes from 1.556 to 1.447.  However, the chances of getting one run across increase from 0.643 to 0.698.  It is possible to play for 1 run as opposed to maximizing total runs scored in an inning.  And sometimes playing for 1 run makes sense (think 2004 ALCS).

 
I think this is the important point.  If Farrell thinks (and/or the Sox think) one run wins a game, he's more interested in the distribution of run expectancies, not the average run expectancy.  Given that the average outcome is relatively close and you're more likely to get a long tail (i.e. lots of runs) with a non-bunt,  the expectancy of one run (and possibly two) is higher by bunting (as lexrageorge mentions above).  The Sox probably have done studies on how many runs it takes to win in late and close situations and decided that despite the lesser overall run expectancy of bunting, the probability of winning is higher because you don't need 5 runs you need 1-2 (on average).  Does anyone know if Farrell bunts more in late and close than early/far?  I couldn't see that data anywhere. 

 

[Hee Sox Choi]

However, the chances of getting one run across increase from 0.643 to 0.698.
 
So you are telling me that we get a FIVE PERCENT boost for wasting an out and bunting the guy over ONLY IF WE SUCCEED with the sac bunt?  So if you include all the times we DON'T get the bunt down, or foul off a couple and then the guy strikes out, or we bunt but the guy gets thrown out at 2nd, etc., I don't see how you can EVER justify bunting.  A 5% BOOST if the bunt is laid down correctly????  FIVE????  That's it?  And with all the other outcomes not included, the attempted bunt is even worse than I ever imagined.
 
My God, why would you EVER bunt?  What am I missing here?

 

[Reverend]

However, the chances of getting one run across increase from 0.643 to 0.698.
 
So you are telling me that we get a FIVE PERCENT boost for wasting an out and bunting the guy over ONLY IF WE SUCCEED with the sac bunt?  So if you include all the times we DON'T get the bunt down, or foul off a couple and then the guy strikes out, or we bunt but the guy gets thrown out at 2nd, etc., I don't see how you can EVER justify bunting.  A 5% BOOST if the bunt is laid down correctly????  FIVE????  That's it?  And with all the other outcomes not included, the attempted bunt is even worse than I ever imagined.
 
My God, why would you EVER bunt?  What am I missing here?

 
There's been some confusion in this thread because there were two bunts last night. Those numbers are for the bunt that moved the runners from 1st and 2nd with no outs to 2nd and 3rd with one out.
 
For moving the runner from 1st with no outs to 2nd with one out it's 0.441 to 0.418.

 

[smastroyin]

The proper way to think about it is to think about all of the scenarios.
 
.643 is the chance with first and second and none out.
 
Now you have a batter.  Several things can happen as a result of this at bat.  If we are talking only about the need to score at least 1 run:
 
Batter reaches safely while a runner scores.  Obviously this increases the chances to 1.000 
Batter reaches safely while all runners advance exactly one base.  This increases the chances to .877 (bases loaded, no out)
Batter makes a productive out that advances both runners.  This is the .698
Batter makes a productive out that advances one runner.  This is .652 (runners on the corners with one out)
Batter makes a non-productive out.  This reduces the chances to .429 (runners on first and second, one out)
Batter hits into DP that takes middle runner.  This reduces to .270 (runner on third, two out)
Batter hits into DP that takes lead runner.  This reduces to .230 (runner on 2nd, two out)
Batter hits into DP that takes both runners.  This reduces to .135 (runner on 1st, two out)
Batter hits into TP that ends inning.  This reduces to 0.
 
I don't feel like doing all the math, so let's presume there are only three swing-away scenarios and that the bunt is successful.
 
Batter reaches safely and a runner scores.  Let's say the chance of this is .200
Batter reaches and no runner scores.  Let's say the chance of this is .200
Batter makes a non-productive out.  So the chance of this is .600
 
.2 * (1) + .2 * (.877) + .6 * (.429) = .6328
 
So the expected outcome of letting the man swing away is a 63.3% chance of scoring at least a single run.
 
The bunt puts you at a 69.8% chance, so purely by the numbers in this case, you bunt.
 
However, if you only successfully bunt at a 70% rate, the bunt is really only worth
 
.7 (.698) + .3 (.429) = 61.7%

 

[Reverend]

We've had at least a couple of instances of the lead runner being thrown out too, which worsens things still.

 

[geoduck no quahog]

The proper way to think about it is to think about all of the scenarios.
 
.643 is the chance with first and second and none out.
 
Now you have a batter.  Several things can happen as a result of this at bat.  If we are talking only about the need to score at least 1 run:
 
Batter reaches safely while a runner scores.  Obviously this increases the chances to 1.000 
Batter reaches safely while all runners advance exactly one base.  This increases the chances to .877 (bases loaded, no out)
Batter makes a productive out that advances both runners.  This is the .698
Batter makes a productive out that advances one runner.  This is .652 (runners on the corners with one out)
Batter makes a non-productive out.  This reduces the chances to .429 (runners on first and second, one out)
Batter hits into DP that takes middle runner.  This reduces to .270 (runner on third, two out)
Batter hits into DP that takes lead runner.  This reduces to .230 (runner on 2nd, two out)
Batter hits into DP that takes both runners.  This reduces to .135 (runner on 1st, two out)
Batter hits into TP that ends inning.  This reduces to 0.
 
I don't feel like doing all the math, so let's presume there are only three swing-away scenarios and that the bunt is successful.
 
Batter reaches safely and a runner scores.  Let's say the chance of this is .200
Batter reaches and no runner scores.  Let's say the chance of this is .200
Batter makes a non-productive out.  So the chance of this is .600
 
.2 * (1) + .2 * (.877) + .6 * (.429) = .6328
 
So the expected outcome of letting the man swing away is a 63.3% chance of scoring at least a single run.
 
The bunt puts you at a 69.8% chance, so purely by the numbers in this case, you bunt.
 
However, if you only successfully bunt at a 70% rate, the bunt is really only worth
 
.7 (.698) + .3 (.429) = 61.7%

 
Finally, an analysis I understand.
 
What it comes down to (and I think even the sabrematicians are talking about minimal changes in odds) is the game situation, which many bring up, and the batters/pitchers involved.
 
I don't see how anyone can assess that scientifically. For example - odds must change if you have an extreme ground ball pitcher, a strikeout pitcher, an extreme ground ball hitter, a lousy hitter, extremely slow base runners, or a hitter followed by a string of lousy hitters...
 
These things change the calculus and rely on a manager's feel versus statistics. Of course if the batter can't effectively bunt, that changes everything.
 
Baseball's a crazy game because non-productive outs are the most likely outcome in any scenario. 
 
One more thing - getting a runner to third with less than two outs may also impact the pitcher's options (curveball/splitter).
 
And one more thing...forcing the infield to play in (versus playing for the double play) should also have an impact - which I guess is accounted for in the scenarios above..

 

[joe dokes]

We've had at least a couple of instances of the lead runner being thrown out too, which worsens things still.

 
Where does the ability of the particular batter to bunt (vs. the teams overall success) and SSS come into play?  If a "good hitter" goes 0-10, he's in a bit of a slump, but the team still thinks he's "good hitter."  At some level, the Sox probably have some kind of idea who they think is a "good bunter."  But is screwing up 4 bunts in a row something that should change that perception?
 
To be sure, I'm not talking strategy here, other than to the extent the discussion is folding "execution" into the mix to further call into question the usefulness of the strategy. 

 

[MyDaughterLovesTomGordon]

 
However, if you only successfully bunt at a 70% rate, the bunt is really only worth
 
.7 (.698) + .3 (.429) = 61.7%

 
Unfortunately, the Sox are at 44 percent right now, so .44 (.698) + .56 (.429) = 54.7%
 
(And, yes, I know that there's no actual reason to expect them continue at that way below average rate, and maybe last night there's reason to expect reversion to the mean, but when you look at the actual numbers it just gets more and more frustrating.)

 

[absintheofmalaise]

I'm not sure if this will help y'all out or not, but here is some information I pulled from Fangraphs that shows sac hits and bunt hits and bunt hit attempts. BU is bunts and BUH bunt hits.
 
[tablegrid= Sacs and bunts ] G AB PA H 1B R RBI BB IBB SO HBP SF SH GDP SB AVG BU BUH BUH% Jonathan Herrera 5 2 5 2 2 0 3 0 0 0 0 0 3 0 0 1 5 2 40.00% Jackie Bradley Jr 4 3 4 1 1 0 0 0 0 0 0 0 1 0 0 0.333 4 1 25.00% A.J. Pierzynski 2 1 2 0 0 0 0 0 0 0 0 0 1 0 0 0 2 0 0.00% David Ross 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0.00% Shane Victorino 3 2 3 0 0 0 0 0 0 0 0 0 1 0 0 0 3 0 0.00% Stephen Drew 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0.00% Brock Holt 2 1 2 0 0 0 0 0 0 0 0 0 1 0 0 0 2 0 0.00% [/tablegrid]

 

[Reverend]

 
Where does the ability of the particular batter to bunt (vs. the teams overall success) and SSS come into play?  If a "good hitter" goes 0-10, he's in a bit of a slump, but the team still thinks he's "good hitter."  At some level, the Sox probably have some kind of idea who they think is a "good bunter."  But is screwing up 4 bunts in a row something that should change that perception?
 
To be sure, I'm not talking strategy here, other than to the extent the discussion is folding "execution" into the mix to further call into question the usefulness of the strategy. 

 
It seems clear that there are some players who are very good bunters. After that, though, everything I've seen gets really hazy with respect to players who are so-so. I also have never gotten a strong sense form what I've read that anyone knows who's better and worse as a bunter outside of the really great ones.
 
But yeah, in theory, you would have to bunt enough to get a really good idea and you might not want to give away that many outs to learn that some of your players can't bunt for crap and even if it means you also learn which guy can bunt really well. It's an interesting case where bunting might have value, but the cost of the information of who has that value is to high to pay.

 

[pokey_reese]

The proper way to think about it is to think about all of the scenarios.
 
.643 is the chance with first and second and none out.
 
Now you have a batter.  Several things can happen as a result of this at bat.  If we are talking only about the need to score at least 1 run:
 
Batter reaches safely while a runner scores.  Obviously this increases the chances to 1.000 
Batter reaches safely while all runners advance exactly one base.  This increases the chances to .877 (bases loaded, no out)
Batter makes a productive out that advances both runners.  This is the .698
Batter makes a productive out that advances one runner.  This is .652 (runners on the corners with one out)
Batter makes a non-productive out.  This reduces the chances to .429 (runners on first and second, one out)
Batter hits into DP that takes middle runner.  This reduces to .270 (runner on third, two out)
Batter hits into DP that takes lead runner.  This reduces to .230 (runner on 2nd, two out)
Batter hits into DP that takes both runners.  This reduces to .135 (runner on 1st, two out)
Batter hits into TP that ends inning.  This reduces to 0.
 
I don't feel like doing all the math, so let's presume there are only three swing-away scenarios and that the bunt is successful.
 
Batter reaches safely and a runner scores.  Let's say the chance of this is .200
Batter reaches and no runner scores.  Let's say the chance of this is .200
Batter makes a non-productive out.  So the chance of this is .600
 
.2 * (1) + .2 * (.877) + .6 * (.429) = .6328
 
So the expected outcome of letting the man swing away is a 63.3% chance of scoring at least a single run.
 
The bunt puts you at a 69.8% chance, so purely by the numbers in this case, you bunt.
 
However, if you only successfully bunt at a 70% rate, the bunt is really only worth
 
.7 (.698) + .3 (.429) = 61.7%

I'm glad that I scrolled down far enough, because I was about to do a lot of this.  I think that we can change the expected value a little though, and tweak these probabilities a bit.
 
I think that one important thing would be including the fourth outcome, batter makes a productive out (which is the same as the intended outcome of the bunt) as one of the possible outcomes.
 
Looking at the last two years of data from Fangraphs, it looks like we get:
35% chance of non-productive out (K, IFF, etc.)
33% chance of productive out
32% no out, which is split into: 20% no out, no score, and 12% no out score (this is looking at hit type distribution (frequency of doubles, triples, HR, etc. with the man on second scoring on everything but a single automatically, and scoring on a single at the average rate which is almost 60%)
 
So it becomes:
.12 * (1) + .2 * (.877) + .35 * (.429) + .33 * (.698) = 67.5%
 
So now, that means that unless you think you can get the bunt executed properly 90% of the time or more (which gives you the 67% chance), you are better off swinging away.
 
AND, that's assuming that you only need one run (so it's the bottom of the 9th and you are at home), because swinging away has a lot more possible outcomes where you can score multiple runs than the bunt does.  So basically, unless you have a perfect bunter, swing away.
 
Note:  Using league averages for hit type distribution doesn't take into account a guy who is a pure singles hitter, which is primarily when you would bunt, but Drew has some good pop, and slugs at least league average. 

 

[smastroyin]

thanks pokey, my example was to illustrate the analysis only, as I noted, I didn't really feel like doing more of the math.  Thanks for doing the math.
 
To be really proper though, DPs, which are the biggest negative outcome, should be included.  I left them out in my illustration but if we were really driving toward a decision, you would, for instance, always bunt with the proverbial guy who only hit hard ground balls.

 

[DennyDoyle'sBoil]

 
For moving the runner from 1st with no outs to 2nd with one out it's 0.441 to 0.418.

 
I never noticed this before.  So the bunt with a man on first, even if successful, not only lowers the the average run expectancy but also the chances of scoring at all?
 
Are there win expectancy tables like the run expectancy tables?  (Or are they all calculated from the same data?)  I mean, does the WE really go down when the home team successfully bunts a guy over from 1st to second in the 9th or later of a tie game?  That's awesome!
 
I guess personnel plays a significant role in all this too, though, as noted.  Bunting AJP over is different from bunting a fast runner over.  Also, for example, the drop in run average run expectancy is irrelevant for the home team in a tie game in the 9th or later, since one run is as good a 1,000.  But it does matter for the visitors, since one run may not be enough to win the game.  Just looking at the tables, you'd think that for the visitors, any significant drop in average run expectancy for the sake of trying to increase the odds of scoring one run would always be a bad idea.  The average run expectancy at the start of any inning is .544.  In other words, even if you get the one run, 54.4 percent of the time, the opposing team is going to tie it up or better anyway.  But, then, a one-run lead with Koji warm and ready is really worth way more than is a one-run lead with Mujica ready to come in.

 

[Al Zarilla]

 
Is there an argument that having the batter bunt makes him pay more attention to watching the ball, everything about it, and doing that can be quickly rewarded with contact? Thus, the next at-bat or the next day, he's had a good habit reinforced?  I know I often read a batter comes out of a slump and says he's seeing the ball better. Is calling for a bunt a way to speed up the process of seeing the ball better?

Bunting and hitting a baseball are such different athletic actions that neither probably helps the other much. Sure, seeing and gauging a 95 mph fastball, whether hitting or bunting, the more of it you experience, the better. Still, hitting involves feet, legs, hips, arms, hands, eyes, etc. Bunting is sometimes called "catching the ball with the bat", in other words trying to deaden the ball so as not to tap it back to the pitcher or corner infielder. Pretty simple action.
 
I keep harping on spring training as a place to practice bunting. Like, they are down in Fla for, what, 5 weeks. What the he!! else do they have to do? Well, lots of other things, but there is still plenty of time to learn or hone bunting. Why are the Red Sox so lousy at it, going back to the Tito era through today?

 

[koufax37]

However, the chances of getting one run across increase from 0.643 to 0.698.
 
So you are telling me that we get a FIVE PERCENT boost for wasting an out and bunting the guy over ONLY IF WE SUCCEED with the sac bunt?  So if you include all the times we DON'T get the bunt down, or foul off a couple and then the guy strikes out, or we bunt but the guy gets thrown out at 2nd, etc., I don't see how you can EVER justify bunting.  A 5% BOOST if the bunt is laid down correctly????  FIVE????  That's it?  And with all the other outcomes not included, the attempted bunt is even worse than I ever imagined.
 
My God, why would you EVER bunt?  What am I missing here?

 
Don't forget two pieces of data missing from this:
 
1) Sometimes the bunt goes better than a sacrifice (reach on hit, or misplayed for error and extra bases).  The 2001 Diamondbacks would like to show you that sometimes this happens, although I would imagine a relatively insignificant portion of the time when Jon Lester isn't involved.
 
2) The run expectancy and probability tables BOTH take into account the situations that got from where they were to the runs scored via bunts.  They are sometimes interpreted as being more absolute than that, but 1st/2nd 0 out run probability includes all the times that the bunt was the decision, runners got to 2nd and 3rd, and then the next guy singled.  So it isn't quite the run probability/expectancy if you choose not to bunt.
 
I'm not sure either of these matters tremendously, and I am almost always against the sac bunt including many of Farrell's recent ones.  But the situation is more complicated than just the run expectancy and probability tables.

 

[koufax37]

Bunting and hitting a baseball are such different athletic actions that neither probably helps the other much. Sure, seeing and gauging a 95 mph fastball, whether hitting or bunting, the more of it you experience, the better. Still, hitting involves feet, legs, hips, arms, hands, eyes, etc. Bunting is sometimes called "catching the ball with the bat", in other words trying to deaden the ball so as not to tap it back to the pitcher or corner infielder. Pretty simple action.
 
I keep harping on spring training as a place to practice bunting. Like, they are down in Fla for, what, 5 weeks. What the he!! else do they have to do? Well, lots of other things, but there is still plenty of time to learn or hone bunting. Why are the Red Sox so lousy at it, going back to the Tito era through today?

 
In addition to Spring Training, there is plenty of time in the busy life of a MLB player to find an extra five minutes here and there for a skill activity that isn't physically taxing and does not affect normal recovery cycles.  As much as I am not a fan of sacrifice bunting, you should either be competent at it or spend 20 minutes a week trying to get at least a little better.  Especially as shifting becomes more common, basic bunting competency will be a more valuable skill for almost all players.

 

[lexrageorge]

I'm glad that I scrolled down far enough, because I was about to do a lot of this.  I think that we can change the expected value a little though, and tweak these probabilities a bit.
 
I think that one important thing would be including the fourth outcome, batter makes a productive out (which is the same as the intended outcome of the bunt) as one of the possible outcomes.
 
Looking at the last two years of data from Fangraphs, it looks like we get:
35% chance of non-productive out (K, IFF, etc.)
33% chance of productive out
32% no out, which is split into: 20% no out, no score, and 12% no out score (this is looking at hit type distribution (frequency of doubles, triples, HR, etc. with the man on second scoring on everything but a single automatically, and scoring on a single at the average rate which is almost 60%)
 
So it becomes:
.12 * (1) + .2 * (.877) + .35 * (.429) + .33 * (.698) = 67.5%
 
So now, that means that unless you think you can get the bunt executed properly 90% of the time or more (which gives you the 67% chance), you are better off swinging away.
 
AND, that's assuming that you only need one run (so it's the bottom of the 9th and you are at home), because swinging away has a lot more possible outcomes where you can score multiple runs than the bunt does.  So basically, unless you have a perfect bunter, swing away.
 
Note:  Using league averages for hit type distribution doesn't take into account a guy who is a pure singles hitter, which is primarily when you would bunt, but Drew has some good pop, and slugs at least league average. 

I don't get the 90% success rate assumption.  
 
I would argue that your equation assumes a 65% success rate (35% chance of non-productive out, which I assume includes unsuccessful bunt attempts).  

 

[OttoC]

One of the problems with sacrifice bunting nowadays is a lot of players try to disguise their bunt intentions (with a lot of those hoping to get a base hit) with teh results that they aren't set up properly to bunt (contrast them to NL pitchers bunting).