Why I Don’t Like Sacrifice Bunts

Put ever so simply, I don’t like sacrifice bunts because by giving up an out to move a runner over one base, the best case scenario is an increased likelihood of scoring less runs. In other words, by attempting a sacrifice bunt, so much can go wrong, while even if it does go right it’s not necessarily to the batting team’s advantage.

Let’s take a look at the bottom of the seventh inning of yesterday’s Minnesota Twins and Toronto Blue Jays game, and compare it to a run expectancy chart that has measured every at bat in every situation from 1993 to 2010.  The Blue Jays were down 3-2 to the Twins with the eight, nine and lead off hitter due up.

  • None on and none out. Run expectancy = 0.539.
  • J.P. Arencibia walks. Run expectancy = 0.929.
  • M. McCoy walks, J.P. Arencibia to second. Run expectancy = 1.542.
  • Y. Escobar sacrifice bunts, J.P. Arencibia to third, M. McCoy to second. Run expectancy = 1.438.
  • J. Rivera grounds out. Run expectancy = 0.604.
  • J. Bautista lines out. Run expectancy = 0.

What the run expectancy number is representing is every time that that situation has arose in an inning over the last seventeen years of baseball, the result has been x number of runs.  So, more runs have been scored in an inning where there are men on first and second with none out than there have been with men on second and third with one out.

Now, I recognize that this is very much a neutral statistic and it doesn’t account for who is batting and who is pitching. For instance, a sacrifice bunt might be the best route if your ninth hitter is a pitcher or just a generally terrible batter. In this scenario though, the person being asked to sacrifice was the lead off hitter, the one person on the team whose sole mission it should be to get on base. If a manager doesn’t have faith in his lead off hitter’s ability to get on base, I’d suggest that the manager hasn’t created his batting order properly or he’s contradicting himself.

Here’s a quick chart to give you an idea of the run expectancy for certain situations:

Run Expectancy Matrix, 1993-2010
BASES 0 OUTS 1 OUT 2 OUTS
___ 0.539 0.289 0.111
1__ 0.929 0.555 0.240
_2_ 1.172 0.714 0.342
__3 1.444 0.984 0.373
12_ 1.542 0.948 0.464
1_3 1.844 1.204 0.512
_23 2.047 1.438 0.604
123 2.381 1.620 0.798

Comments (65)

  1. Go put that chart in binder like Joe Girardi nerd!

  2. Checked out the game log on FanGraphs this morning for the same inning and Escobar’s sacrifice was awfully close to being neutral in term’s of win probability added, it checked in at a -.001 and was the most productive, or least harmful if you will, out of the game.

    That being said, yea, still not a good idea, and really hope Farrell isn’t going to role this tactic out all that often in the future. Really disappointing to see any manager do this sort of stuff.

    And they also got lucky too that Escobar didn’t screw it up and get Arencibia thrown out on a force play at third, he’s not too a fleet of foot. When the play works as intended and is still a net negative result for your chances of winning the game, that should be a sign.

  3. That chart is probably the opposite of what Girardi does. He minimizes situations to the point where the sample size becomes so minuscule his findings are useless. Think more like: When Brett Gardner faces a left handed pitcher in the seventh inning or later on a Wednesday night that has an odd numbered date, he’s 2 for 4, so I’ll let him hit.

    • I have to say I don’t mind the bunt in this situation. They were playing for one run which, while flawed thinking in its own right, isn’t a bad strategy in the end game situation.

      Juan Rivera hitting behind him renders it all moot. His chances of reaching base is pretty much zero. He’s the worst.

  4. I haven’t looked it up, but my impression is we’ve seen more bunting this past weekend than we saw under Cito’s last two seasons combined. That said, I agree it was odd having Yunel bunt. Though he is one of the more experienced bunters on the team, I agree he should have tried for a proper hit which could have stood a better chance of an actual RBI.

  5. And they got unlucky that Escobar had to recoil from the pitch and didn’t outrun the throw to 1st. You can’t treat sacrifice bunts like guaranteed outs, especially with someone as fast as Escobar dropping it. If there’s any hesitation because they think they can get another runner, or they try for another runner and fail, you have bases loaded with no one out, which is a BIG increase in run expectancy.

    The bunt may not have been completely the right call, and it didn’t work, but it was certainly defensible and definitely not the main reason they lost.

  6. Completely agree for the most part.
    Obviously those numbers will be affected by big scoring innings though. So yes, more runs have been scored when runners are on first and second. But what if it’s 3-3 in the bottom of the 9th? You only need one run, more than that is useless. Whats the percentage of just “runs” scoring (be it 1 or 6) between the two situations. I’ll take a runner on 3rd with one out (0.984) over runners at 1st and 2nd with no outs (1.542).

    But only in certain situations.

  7. Being one run down seems to me like even more of a reason not to bunt. You’ve already got a guy in scoring position. You’re at home. Who’s more likely to get on base: Escobar or Rivera?

  8. I really liked the play. I think Escobar has a better chance of turning a sacrifice bunt into a base hit than allowing them to get a force at 3rd Fangraphs only had it as a -.001 play, so I think on a whole your looking at a positive win expectancy play.

  9. Do stolen bases next. The negatives far outweigh the positives.

  10. Just because it’s a neutral overall situation doesn’t justify doing it. In fact, if it doesn’t really change the run expectancy, that’s an argument against doing it as so much can go wrong.

    Saying that because Escobar has the ability to hit a bunt single, doesn’t make it better either. There are many more positive outcomes that can come out of him swining freely as his expected OBP is much higher, not to mention the fact that he can hit a double, triple or homerun as well.

    Sacrifice bunts aren’t completely useless, but they’re pretty damn close. There are only a few situations where it makes sense and managers rarely deploy it in those circumstances.

  11. What I think EthanDR is trying to bring up, is that run expectancy does not provide the whole picture. If you are only playing for one run, you need to look at the probability of scoring 1 run in each situation rather than looking at the expected number of runs in each situation. Obviously, as Parkes just brought up, letting Escobar hit may still be the optimal play with a runner already in scoring position, but basing your decision purely on expected runs in this situation may be flawed.

  12. The point I made on Twitter, during the game, was that the move required perfect execution to lower your run expectancy. It’s such backwards thinking.

  13. If them bunting was an indication they were playing for one run, when is that ever a good idea in the seventh inning of a home game with the top of the lineup coming up? Why would you want your team trying t score once in the seventh to accomplish nothing more than tying the game? This isn’t football(soccer), there’s no draws.

  14. Like I say in the post, it’s a neutral stat, and you definitely can’t let it be the sole basis for your decision, but it still acts as a good piece of guidance.

    Good points from Tom and Cam, why on earth do you play for one run when you have a chance at more? Especially so, when playing for one run makes scoring that run less likely than if you were to play for two or more.

    And Travis is right too. If it’s a barely negative outcome, why would you try to do that? The fact that your leadoff hitter is up to bat makes it even worse. I have no problem with a fast guy trying to get a bunt hit, but that wasn’t what was happening in this play.

  15. I was cool with the bunt and here’s why:

    While your graph may suggest that it’s statistically prudent we swing away (although 1.54 and 1.44 are very close), that graph doesn’t consider how often a single run is scored in that situation, compared to that of multiple run innings. Barry Sanders had an amazing yards per carry average but your not running him up the gut on 4th and 1 because some graph tells you he’s going to get you 6 yards.

    At that point in the game you NEED 1 run. You use arguably your best bunter to sacrifice both runners into scoring position (he was a half step from being safe btw) and afford your 2 hitter (the guy who is supposed to be the most adept at situational hitting) the luxury of hitting a fly ball or hitting it on the ground to anyone but the pitcher, instead of asking him to get a base hit. Rivera, incidentally hit a ground ball back to the pitcher, which was unlucky in itself, although don’t expect me to advocate for Juan to spend alot of time hitting second this year.

    If it’s the 6th inning or earlier, I can bet you dollars to donuts Farrell let’s Yunel hack and plays for 2+ run inning.

    That .10 doesn’t take the situation into account Dustin, as you yourself mentioned above. If you’ve ever played in or managed a tight ballgame you’d give that point a little more credence. It was ABSOLUTELY the correct decision to let him bunt and I dont think anyone in that clubhouse believes otherwise.

    That being said very positive opening series, got a little unlucky there at the end that game could have easily gone our way. Nathan looked horrendous.

    • I still don’t understand why you’d play for one run when that decreases the likelihood of scoring one run.

      Run expectancy is related to win expectancy. In both cases, it resulted in small dropoffs in probability that the batting team would score or win.

  16. What JRock said. Shouldn’t this analysis be based on win expectancy rather than run expectancy? The goal of the game, after all, is not to score as many runs as possible, but to win. Not that it would have changed the optimal strategy in this case, as Tom Pinzone already pointed out. But in a situation like tie game, runner on 2nd, no outs in the 9th, run vs. win expectancy would probably call for different strategies.

  17. “I still don’t understand why you’d play for one run when that decreases the likelihood of scoring one run.”

    I’m not sure that it does. I don’t know if this is actually true, but I can imagine a scenario like this:

    strategy 1: RE 1 run
    strategy 2: RE 1.3 runs

    where the REs are based on the following:

    strategy 1: 20% * 0 runs + 60% * 1 run + 20% * 2 runs = 1
    strategy 2: 10% * 0 runs + 40% * 1 run + 30% * 2 runs + 10% * 3 runs = 1.3

    The first strategy has a lower RE but a higher probability of scoring 1 run. Again, I don’t know if this actually happens, I just made up the figures, but it seems intuitively like it could be true in some cases.

  18. @Xave

    No, because when ever you use wins (even team wins) as the stat to determine whether a given offensive strategy is a sensible one you are giving that strategy credit for half of the game that it has no effect on whatsoever: run prevention (pitching a defense). Whether you bunt or do not bunt does not have an effect on how many runs you do or do not allow going forward.

    Also, there is a chart in “Baseball Between the Numbers: Why Everything You Know About the Game Is Wrong” that shows your percentage chance of scoring one run as well which is useful to look at in the position some of you are describing.

    Runners / 0 Out / 1 Out / 2 Out
    Empty / 28.0% / 16.5% / 7.1%
    1st / 41.7% / 27.2% / 12.7%
    2nd / 62.5% / 41.0% / 22.9%
    3rd / 82.7% / 66.1% / 25.4%
    1st & 2nd / 61.6% / 41.4% / 22.8%
    1st & 3rd / 84.6% / 64.5% / 26.8%
    2nd & 3rd / 86.1% / 67.4% / 26.6%
    Loaded / 85.6% / 65.4% / 30.7%

    So, as you can see, your chances of scoring just that one needed run (if that is indeed all you need and you are willing to give up some fraction of a potential expected run to try and incrase your odds at getting it) are 61.6% with runners on first and second and nobody out. If you successfully sacrifice them over then you are increasing your odds at scoring that one run to 67.4%. You are taking away the chance of getting something better entirely (maybe the hitter sacrificing would be the one to come through with the game winning hit otherwise) and taking the risk that he pops up the bunt and you get nothing for a 5.8% better chance at scoring one run. I’m not saying that it is never a good idea to do that, but I think that the circumstances under which it is the right strategy are much more limited than most people seem to think.

  19. I do not think run expectancy and run likelihood are equivalent. Are you referring to likelihood in common usage, i.e. synonym for probability, or in the statistical sense?

  20. They are related but not twins. (*made up unrealistic numbers but prove point) you could have a situation where if you bunt you score 2 run 4/4 times (run expectancy of 2) and a situation where if you don’t bunt you score 5 runs 2 times and 0 runs 2 times. (run expectancy of 2.5) In the bottom of the 9th (or i would go out on a limb and say 8th) down by 1 run the bunt is a lesser run expectancy but will have a higher win expectancy and is therefore the right managerial decision.

  21. @Nick Thanks for the probabilities. That makes the case for not bunting right there.

  22. call up Tony Fernandez, we need to bring back the Baltimore Chop.

  23. @Nick can’t have your cake and eat it too. Can’t say that bunting is taking away probability of bunter doing something without saying bunter could strike out and bring probability down to 41.4% for one run. The fact that he could have that hit is already included in the 61.6% with no outs and runners on 1st and 2nd.

  24. Great stuff, Nick. Thanks.

  25. I think there’s a better chance that Escobar reaches than Arencibia gets thrown out. So if the win expectancy was neutral with what actually occurred, then if you account for the fact it could have been a bunt single or a fielders choice without actually getting the runner, and say that had a greater chance than getting Arencibia at third, you would have probably had a better win expectancy going in to the at bat bunting rather than if Arencibia is swinging.

    That said, I also agree with Parkes, etc. saying that there is no use only going for one run. Both for the reason that a tie isn’t good enough and there’s still a lot of game left where you may need those extra runs (like in case Jon Rausch gives up a long home run to a speedy, low power leadoff hitter).

  26. Thanks for the probabilities Nick! Will have to check out the book. I must point out, of course, that in the scenario I described bunting is the right strategy, increasing the likelihood of one run from 62.5% to 66.1% ;) Though that seems to be the only such case…

  27. I meant Escobar swinging. An edit button would be nice for comments.

  28. Interesting argument, some good ideas. The one thing that often pisses me off about bunting, whether it be with 1 or 2 on, is that you give yourself up to get a guy across… and then the pitcher just tosses the unintentional intentional walk. Sure, you’ve got 2 on now, but you’re still a ground ball away from the inning being over.

  29. @Jack Keeler

    You are right of course. The bunter could strike out or ground out or pop out or hit a screaming line drive and have it go right at somebody. My point is twofold…

    1) If your bunt is successful you have increased your chances at scoring that one needed run (assuming that playing for one run is the right move) by 5.8% in the situation we are talking about.
    2) Bunts do not have a 100% success rate. Sometimes the bunt goes right to the pitcher really quickly and the result is a double play making the new situation a runner on third with two outs (a 25.4% chance of scoring that one run that your team needs so badly, 36.2% lower than the situation you were just in). Sometimes the bunt is popped up and then you have given away an out for nothing (lowering your chances of scoring to 41.4%, 20.2% lower than your chances before the attempted bunt) or, if the runners got great jumps you may have even popped a bunt into a double play if one of them doesn’t get back quickly enough and the catcher makes a great throw.

    If bunts had a 100% success rate then of course it would be the right move when playing for one run (assuming you aren’t taking the bat out of a great hitter’s hands and giving it to a greatly inferior hitter), but they do not. There is a non-trivial chance that the bunt will fail to have the desired effect and then you have given away one or more outs for no benefit on the bases.

  30. @Xave. But you need to consider the probability that the sac bunt does not work. If that occurs, then you are down to 44.1%.

  31. Just curious,

    But would a better comparison not be the following:

    first and second, none out and next play is sacrifice bunt vs. first and second, none out and all other next plays.

    Parkes, you are relying on the fact that first and second, none out has a higher run expectancy, but wouldn’t that number include all those situations where managers use a sacrifice bunt. I’m not saying you are not right (because I don’t know the numbers for the comparison I posed) but it seems to me the above comparison would provide better guidance. If the run expectancy for you first-second, none out is higher because 80% of the time in that situation a sacrifice bunt is used, doesn’t that go against the conclusion you are reaching?

    Just a thought – I am really interested to know the answer and am happy to have someone explain to me why I am wrong or how that is already taken into account.

    Really great blog by the way. Thanks for the all the work you do on it.

    - CGB

  32. @JRock It goes both ways – there’s also the chance the runner makes it, in which case you’re up to 84.6%.

  33. I think everyone is missing the point of grouding into a double play, which should be factored in…RE goes to 25.4%. This threat is eliminated with a successful bunt.

  34. Also, Curious George has a great point

  35. As an addendum to my previous comments, the same chapter from the book reveals the following probabilities:

    Situation / Successful Sacrifice / Batter reaches / Failed Sacrifice / Double Play
    1st, 0 out / 72.9% / 13.6% / 11.2% / 2.0%
    2nd, 1 out / 71.6% / 10.6% / 14.4% / 3.2%
    2nd, 0 out / 69.5% / 18.6% / 11.4% / 0.5%
    1st & 2nd, 0 out / 67.0% / 17.1% / 13.4% / 2.2%

    There is the chance that the batter reaches base to consider, of course, but overall if you look at that scoring one run probability table as well as Parkes’ expected run table the damage of a potential failure is sufficiently higher than the benefit of the hitter reaching base to make sacrificing carry more risk than potential “surprise reward”.

    You aren’t, on average, 5.8% better off as a result of attempting a run. Only if the bunt is successful.

  36. Spitballer: Grounding into a double play is included in that 1.542 number.

    The possible scenarios of swinging the bat are all accounted for in that number. The number where there isn’t any probabilities is for the bunt, unless we assume it is successful and results in the batter being out, which isn’t a great assumption.

  37. @Nick Wernham

    This is a good argument however I am going to argue that with a player such as escobar the good and bad are going to cancel each other out. I would think that (perhaps incorrectly) escobar seems to handle the bat well and is a very good bunter. I would say that on a play like that the odds of him getting on base are similar to him not advancing the runners. and for the other niche events (double play) etc. i would say there is a similar probability albeit slightly smaller that the ball is thrown away and runner scores game over.

  38. Basically, I am in favour of bunting if the following things are all true:

    1) It is the right decision to only play for one run.
    2) There are no outs. I never want to see a bunt with one out ever.
    3) The odds of scoring one run go up with a successful bunt.
    4) You are confident in your hitter’s ability to get that bunt down successfully. This is subjective of course, but I think that this is an area where good evaluation of your own players’ individual talents can potentially pay dividends.
    5) The hitter following the bunter is a substantially better offensive player. They are far more likely to be the one to bring home that needed run.
    6) The hitter behind that hitter is an improvement on the bunter as well, in case the team decides to walk the guy who follows the bunter to set up the double play.

    If ANY of those things are not true then you do not bunt.

  39. @Spitballer

    The situation you are describing is part of the tiny minority that makes up the remainder of for each situation in that table. The odds of that are lower than the double play, albeit more rewarding since in this situation we are playing for one run and well, that’s the game.

  40. Im a big fan of bunting if 1 3 and 4 are true

  41. It makes sense to me. I’m not uniformly opposed to “small ball” (I like hit-and-run plays a lot provided the right guys are at the plate and on-base and I’m all for guys working on their baserunning and learning when to take the extra base where to appopriate), but bunting tends to be an overrated tactic simply because it is often an exciting play to watch.

    Using specialized tactics is not a useful enterprise unless those specialized tactics are likely to yield the desired results. We have all of this useful information at our fingertips, why not use it to determine strategy. Ignoring it seems needlessly reckless.

  42. @nick

    did the numbers with the 2 charts you provided you are correct average does not rais 5.8%. The odds of bunting including all of these probabilities ( which ignores probabilities of wild throws) turns out to be 65.6% for 1 run as opposed to 61.6% for 1 run without bunting.

    (.67)(.674)+(.171)(.856)+(.134)(.414)+(.022)(127)=.656

    So I would disagree with you and parkes. I would say you should always bunt if 1 run is going to win you the game unless you have statistically significant future reasons not too. (bunting with jose to bring up j-mac)

  43. Nick, your numbers don’t take into consideration who is at the plate. Often, managers will ask whoever is at the plate to lay a bunt down and it doesn’t always work. It’s not as simple as in all situations where this has occured outcome A has happened X amount of times, outcome B has happened Y amount of times and C happened Z amount of time. That doesn’t take into effect when a pitcher who can’t handle the bat lays a bunt down, when a catcher with no speed does it, etc.

    The bunt was the right move at that point in the game. You’re not playing for one run. You’re putting two runners into scoring position, and scoring one from third if Rivera can get one into the outfield for a sac fly. No, it didn’t work, but the same can be said if Escobar grounds out. Let’s assume the inning plays out the same except for the Escobar at bat:

    Batters get on base approx. 3.5 times out of 10. That leaves 6.5 times out of 10 where Escobar gets out. Of those other 6.5 times let’s assume he can move the runner over on a sac fly to right (that’s about the only way I can think of). Mccoy then stays at first. Rivera grounds to the pitcher, double play, inning over. Yes, run expectancy may have gone down slightly (about .1 runs), but what are the double play expectancies with runners at 1st/2nd no out vs 1st/2nd 1 out vs. 1st/3rd 1 out vs. 2nd and 3rd 1 out. I’d be willing to bet the last is the lowest expectancy for double plays. They were playing for a hit from Rivera to score two, would have been happy with one on the sac fly, and it just didn’t work out that way. Doesn’t make the sac bunt the wrong play.

  44. Came on here to defend the bunt, but it’s been covered pretty thoroughly. My main point was Jack Keeler’s, that run expectancy is an average and while the bunt cuts into the big inning potential putting a runner on third with one out as opposed to second with no outs makes THAT run more likely to score. And the value of the tying run shouldn’t be understated – sure you have to get another run eventually, but tying the game gives a potentially infinite number of innings to score it, while if you’re trailing by a run you have a fixed number of outs (9) to play with.

    I’d also be interested in the percentage of errors on all bunts vs. every other play. I imagine the fact that fielding the bunt is a play in which the fielder is in motion leads to more balls thrown away – you’re less likely to have a lazy fly ball or routine grounder on a bunt.

  45. Good discussion. Another angle I’ve been pondering…
    Has anyone heard that the Jay’s are more interested in using speed this season? No? So go crawl back under your [Getting Blanked] rock! For the rest of us…
    We had lot’s of talk & demos pre-season, and now have early execution from BlueJays v2011 of speed (double steal, sac bunt, etc). This, when we all “know” that the stats say speed is often/easily overrated. However, I wonder if the main value of speed is the threat of it, e.g. rushing the “D”, complicating the Managerial decisions. If so, then advertising it makes more sense as does displaying it early. Then, play more conservatively until you see the effect diminishing and repeat above. Pure horsedoodle?

  46. Also, if you aren’t going to sacrifice in the exact situation that arose in the 7th, when can you ever defend the bunt? Given the score, the situation and the fact that there was a good bunter at the plate, I can’t come up with a single more appropriate bunting scenario.

  47. @nick at 1:10 pm and xave
    Win expectancy should be used, not chances of scoring one run. What win expectancy does is gives you a perfectly league average offense and defense fight it out for the top of the ninth (and bottom if needed). This needs to be a part of your thinking as to whether it is worth going for one run or not.

    Also, one thing not contained in any of the above analysis is that defensive strategy and its interaction with offensive strategy is not considered (ie game theory). Defensive positioning has a lot to do with how successful a bunt might be. If a manager never ever calls a sac bunt the defense can play deep and increase their chances of a GIDP. If you always sac bunt the defense can pitch out or start the corners way in etc.

    The ideal strategy hence lies somewhere in between. See http://www.fangraphs.com/blogs/index.php/were-the-yankee-sac-bunts-in-the-8th-inning-correct/ for a good discussion of the topic.

    One of the interesting things about using game theory in this is that the ideal decision would be a random choice between bunt or not bunt (weighted accordingly). eg you might bunt 1 in a million times with Pujols up and 99 in 100 when a pitcher is up.

  48. I should also mention I don’t blindly support Sac Bunts either. When Mike McCoy turned to bunt in three at bats I was not happy. When he showed bunt on the first pitch he saw at the ninth I screamed “It is the bottom of the 9th. We need to score 2 runs. A runner in scoring position will not help right now. DON’T BUNT!” My brother turned to me and said “Well said. I like how you hit each important point there.” So, yeah, a sac bunt is definitely not always the right move.

  49. thats something i think everyone can agree with. if you didnt throw your shoe at the tv when mccoy signaled bunt in the 9th you are a better man than I

  50. That fangraphs article was enlightening. I like the poker analogy. Made me wonder if some of those old-boy managers who tend to gut-manage and forsake the stats (La Russa, Scoscia come to mind) can attribute some of their success to their unpredictability. While a manager who ties himself to stats without understanding them can become dogmatic, the guys who manage by feel are like poker players who don’t bother with the math, just the psychology.

  51. In the 7th inning the Jays we’re going for the lead, which would have been 2 runs. So run expectacy is inaccurate because it only shows the expectancy of 1 run…

    In regards to Mccoy bunting, What you fail to acknowledge is the expectancy of a double play. Managers sacrifice bunt to have a runner move in to scoring position when they need a run, but most importantly you stay out of a double play which would all but kill your run expectancy..

    So its a sacrifice of an out to prevent 2 to be taken with one swing.

  52. dc are you reffering specifically to the 9th inning? Because giving away an out in the 9th, down two makes no sense. Staying out of a double play but putting yourself 50% (!) closer to losing is not worth it. You also don’t reaise your chances of winning the game in that situation, because even if Escobar gets a hit, you still probably need at least two more, with two outs, just to tie it.

  53. Jeff— Not speaking of the ninth, the article refers to the 7th inning and I was commenting on that scenario..

    Runner on first, batter is Mccoy… I bunt everytime, because I have no faith in Mike Mccoy as a batter, had it been nearly anyone else, no bunt.

    Mccoy works a walk, 2 on no one out Escobar batting. . 2 options

    1) allow Esco to swing away, a single would score one run, would need a double to score 2.. GIDP would leave a runner on with 2 away.

    2) Lay down a sac bunt (almost beat it out) have a runner on 2nd and 3rd with one away, requiring only a single to take the lead.

  54. Yeah, that’s ok. I didn’t particularly have a problem with McCoy bunting there, either. I found it funny how he tried on all four pitches and Mijeras couldn’t even find the plate.

  55. @MarkV
    Yeah, that’s what I was arguing for, I think. Seems to me win expectancy would cover all the bases better than run expectancy, including the rare case (which I still defend) where bunting is optimal.

  56. McCoy flashing bunt in the ninth was not a sacrifice – you’d have to be nuts to sacrifice there. He would have clearly been bunting for a hit, unless he blinked and completely missed the Span jack or something.

  57. I’d be interested to see a chart of bunt outcomes with a few constraints: no pitchers in the sample and only regular bunters (min 5/year). I think this would give us a better idea of the utility of the bunt when attempted by those who are presumably competent at it.

  58. interesting discussion, guys. i’ve a couple of thoughts i thought i’d share…

    you can’t just say that a sac bunt lowers run expectation, therefore you shouldn’t do it. any out will lower RE, a GIDP will lower it significantly.

    so the question is “if i let escobar hit normally, what is the run expectancy likely to be after his at-bat. Lets call it “expected run expectancy”, ERE. If the ERE is greater than the run expectancy after a sacrifice bunt, don’t bunt. if it’s less, maybe a bunt is the smart call. this is why intuitively it seems obvious to let the pitcher bunt. but how bad does a batter need to be before a bunt becomes the right call? here’s my attempt at putting some numbers on it…

    to work this out you need to add up the run expectancy of all the possible outcomes, weighted by how likely that batter is to provide that outcome. For example, single:1st and third, no out,one run scored, so (1 run + 1.844 RE)=2.844. Escobar hits singles 19.2% of plate appearences, so add (2.844*0.192 = 0.546)ERE. Add up over all possible outcomes, and see what you get. In this particular situation, i got 1.56: higher than the 1.43 after a sac bunt. so your call was right, don’t bunt. What about johnny mac? his ERE is just 1.40, so maybe bunting is correct here. so the cutoff point is somewhere in the middle. Mike mccoy? i’m not sure, he doesn’t have enough major league at-bats. Aaron hill came in at 1.54, Adam lind at 1.57, EE at 1.53. So you’re right that in most cases it’s best not to bunt, but it’s more subtle than you’d think.

    now, of course if the goal is to score just one (or two) runs, you need to do the same thing with “one-run expectancy” in place of run expectancy, but i don’t have those figures, and i’m all excel-ed out!

  59. one other thing that i don’t think has been mentioned… did farrell make the call to bunt, or did escobar do it on his own? we saw last year that he has some funny ideas about when to bunt, and moving him to leadoff can only stop his shenanigans in the first inning!

  60. ^^Did you not see Butterfield (or another coach, fairly certain it was Butters though) bring Bautista over to directly translate something to Yunel between batters? The call definitely came from the bench.

    Also, people seem to think Parkes is saying don’t bunt and you automatically win there. Well no, that obviously isn’t the case. Yunel could have promptly hit into a double play and Rivera could have K’d right after him and it still would have been the better decision. The goal is to put your team in the best position to succeed. Farrell’s definition of success there was obviously tying the game, unfortunately with two on no out and your leadoff hitter up he should learn to have higher expectations.

  61. Tying or taking the lead. Independently, the two runners on base were exponentially more important than any subsequent runners because the first one ties the game, and the second one gives you a one-run lead. Any possible outcome that involves a home run or a three-run rally is just gravy – those two runs are the game right there. I know the 7th isn’t the 9th, but that late in the game I much prefer a tie or a one-run lead to a one-run deficit.

  62. Personally, I don’t see how run expectancy can be predictive. I could be missing something and if so please feel free to enlighten, but a simple average outcome of 17 years of data devoid of any but the most simple context and ignoring multiple control variables is just a summary of past events and is extremely misleading. Now, 17 years of data showing outcomes with similar base runner, batter, pitcher, defense and park profiles would be predictive. Barring that it amounts to white noise for me. That said, I still don’t like the sacrifice bunt in most situations for the simple fact that the number of outs in a game are finite and giving one away seems counter-intuitive to me.

Leave a Reply

Your email address will not be published. Required fields are marked *