Bunting is normally a bad idea

On Tuesday afternoon, I watched along with some of you in horror as Chad Holbrook bunted Joey Pankake in the first inning after Carolina put the first two runners on.  Holbrook stated ““I wanted to score first today and I told myself that if we get the first two guys on, we’re bunting.”

For many reasons, this was a terrible idea, and bad process led to calamitous results for the Gamecocks.  Pankake bunted right back to the pitcher, who whipped the ball to third and recorded the first out of the inning.  The Gamecocks would exit the inning 0-0, and though they ultimately scored the first run that Holbrook found so important in the top of the second, it (shockingly) was no more relevant than any of the other runs scored in the game.  The missed opportunity in the first, where Carolina went scoreless, wasn’t important because it meant the Gamecocks would be on the board first.  It was important because Carolina needed runs to win the game, and the first inning was an opportunity to score those runs.  Bad coaching and bad execution cost the Gamecocks, and in a one-run game, could have cost them Omaha.

The purpose of this post is to set a standard for when it’s a good or bad idea to bunt.  As Twitter (at least, my account) exploded in anger at the coaching decision, I found myself perturbed that few people were articulating exactly why it was a bad idea.  Many thought it made sense to win or lose with your best players, but I wanted to put together a post that we can use as a reference point in the future to understand when bunting is a good or bad decision.  So let’s look through the data and then analyze the three bunt calls that Holbrook made yesterday, to see how they hold up based on that understanding.

Background

We need to start with this outstanding chart from Boyd Nation, who tracks college baseball over at his website BoydsWorld.com.  Using two years of data from 2011-2012 (after the new bats were phased in), we can understand the expected number of runs for any situation, based on how many runs were ultimately scored by teams in college baseball during this two-year period.  Of course, this is not context-specific, but it’s a point of departure that we can use, and then modify as appropriate to adjust for context.

Below is the expected number of runs for every situation you encounter in an inning, as well as the percentages that you will score at least run (which is useful for late-inning analysis).

Expected Runs
At Least One Run
Outs
0
1
2
Outs
0
1
2
Empty
0.63
0.32
0.12
Empty
32%
18%
7%
1st
1.07
0.63
0.26
1st
49%
32%
15%
2nd
1.39
0.82
0.38
2nd
70%
47%
25%
1st / 2nd
1.79
1.10
0.52
1st / 2nd
71%
49%
27%
3rd
1.67
1.14
0.45
3rd
89%
73%
32%
1st / 3rd
2.06
1.36
0.61
1st / 3rd
89%
70%
34%
2nd / 3rd
2.30
1.56
0.68
2nd / 3rd
89%
73%
32%
Loaded
2.69
1.79
0.89
Loaded
90%
72%
39%

The Situation

So how should we analyze this data?  Typically, it depends on the situation in which the coach is calling a bunt.

Let’s assume for a moment that even Holbrook wouldn’t lay down a bunt with only a runner on first until the late innings (we’ll tackle that analysis later in the post).  At this point, let’s look at how a baseball coach should be looking at the game generally in the first six innings – how can I score as many runs as humanly possible?  In the late innings, I understand valuing the opportunity to score one run above scoring as many runs as possible, because the goal isn’t to win by as many runs as possible, but to win generally.  But before the seventh, that’s just the wrong way to look at it – in those first six innings, you’re working to improve your late-game situation as much as possible, and that means you’re making decisions based on what scores the most (or defensively, prevents the most) runs.

In each of the three situations where Holbrook called a bunt yesterday (Pankake in the first, Saiko in the fifth, and Vergason in the sixth), the Gamecocks had runners on first and second with nobody out.  As we see above, the expected runs in that situation is 1.79 runs per inning, on average.  The goal in each of these situations for Holbrook was to move the runners over to 2nd and 3rd, at the cost of an out.  As the chart shows up, a team with runners on 2nd and 3rd with one out expects to score 1.56 runs per inning.

Let’s make sure we stop here and realize what this means.  In the three calls that Holbrook made, he WANTED this outcome.  Thus, Holbrook affirmatively chose yesterday to reduce the Gamecocks’ expected runs by 0.69 runs ([1.79-1.56]x3).  That’s an incredible run reduction that can be traced to just one thing he did.  That’s inexcusable.

Possible Outcomes

Now, the question that many of you might have right now, and that I also wondered here, was if this accounts for two other potential outcomes: the bunt successfully allows the runner to reach first, or the batter strikes out.  One thing we’re not going to easily be able to do is determine how much two failed bunts harms the batter in his attempt to hit.  The other thing I took off the table is the obviously calamitous result that would be a double play of some sort.  It’s a non-zero possibility, but probably minute enough that we can still smartly look at this question without considering the possibility.

On the other hand, there’s obviously the chance that the opposition commits an error that allows the lead runner to score and puts the other runners on second and third.  So let’s consider those five possible outcomes:

(1) Runners on 2nd and 3rd, 1 out (the theoretically desirable outcome and most likely – successful bunt, successful defense)

(2) Bases loaded, 0 out (the runner reaches base through speed or inept fielding  - successful bunt, bad defense)

(3) Runners on 1st and 2nd, 1 out (the bunter fails, either via strikeout or popout; or, the defense gets the lead out at third – bad bunt, successful defense)

(4) Runners on 2nd and 3rd, 0 outs, 1 run scored (the best-case scenario – the ball is fielded and is thrown away – the runner on 2nd scores, and the other runners land on 2nd and 3rd – successful bunt, failed defense).

(5) Runner on 2nd, 2 outs (the bunter pops up and somehow a runner gets doubled-off – let’s assume for simplicity that the runner is the trail runner, which is charitable to the offense in this analysis – failed bunt/running, successful defense).

John Whittle of The Big Spur graciously offered his best estimate on how often these events occur in the South Carolina games he watches against SEC competition.*  Of course, all of this information struggles because “average” college baseball varies across leagues, so there’s only so much accuracy we can bring to this.

*An important caveat – John did not assist with or endorse this analysis or these conclusions.  His help came from trying to discern how often these events occur, not whether the decision to bunt itself is a good idea, or whether this is the right way to analyze that question.  This analysis is mine, and so any issues with it should be directed squarely at me.

John estimated that the bunt succeeds 70% of the time (2/3, 1 out), the bases end up loaded 5% of the time (1/2/3, 0 out), the bunt fails or the lead runner gets gunned 10% of the time (1/2, 1 out), and the defense airmails the ball so a runner scores and the other runners advance 5% of the time (2/3/0 out).  I didn’t give him the fifth option, but let’s work from this premise, because it’s as good a premise as I can find.

In that scenario, we’d expect the decision to bunt to give away 0.1 runs, all else being even:

Result Exp. Runs Odds Adj. Runs
2/3, 1 O 1.56 70.0% 1.09
Loaded, 0 O 2.69 10.0% 0.27
1/2, 1 O 1.10 15.0% 0.17
2/3, 0 O 3.30 5.0% 0.17
Total 100.0% 1.69
Swing Bunt
1/2 0 O 1.79         1.69

Here, we took the expected run totals and multiplied them by the percentage chance of that result.  By adding up all of those numbers, we can reach the expected value of a bunt (1.69 runs).  Compared to allowing the hitter to simply swing away, that reduces the expected value of runs in an inning.  Over the course of a season, this is a relatively minor loss, but in a one-run game, giving away edges is what differentiates winners from losers.

Another way to look at this is to determine the odds that the player successfully bunts, poorly bunts, or makes a critical error.  You can also estimate the same chances for the defense.  So let’s say the hitter normally bunts successfully 80% of the time, does poorly 10% of the time, and does catastrophically the other 10%.  Let’s assign the same percentages to the defense.  What does this lead to?

Result Exp. Runs Odds Adj. Runs Bunt / Defense
2/3, 1 O 1.56 66.7% 1.04 Successful bunt, successful defense
Loaded, 0 O 2.69 8.3% 0.22 Successful bunt, bad defense
1/2, 1 O 1.10 8.3% 0.09 Bad bunt, successful defense
2/3, 0 O 3.30 8.3% 0.28 Successful bunt, failed defense
2, 2 O 0.38 8.3% 0.03 Failed bunt, successful defense
1/2 0O 1.79 100.0%         1.66

Note, the percentages here are weighted.  If you want the math, ask me on Twitter or via e-mail and I can explain the method.  However, this leads to the same conclusion.  You can tweak the estimates to get different answers, of course, but under what I would consider to be two reasonable sets of assumptions, Holbrook made a poor decision on Tuesday.

Going Further

So let’s step back from the data and assess the specific situations on Tuesday.

1st inning, 0 outs, Pankake

For me, this was the most galling of the three bunts.  Pankake has been a hot hitter lately, and Holbrook immediately made a decision to “score first.”  However, this makes no sense against a team that hits and fields as well as North Carolina, particularly given that Holbrook said on Saturday that “you’re not going to win against North Carolina 2-1 or 3-2. They’re going to put pressure on you. They’re going to score some runs. Every opportunity you get to score yourself is magnified.”

And, he’s right.  Put the analysis aside.  It’s the first inning, there are no outs, and you have your 3-4-5 hitter coming up.  If you think you need crooked numbers to beat the Tar Heels, there will come no better time to get them.  Moreover, it was clear from the first two batters that Moss was rattled, and even from his first pitch to Pankake, which was wild high and tight.  The bunt not only bailed Moss out from a big inning, but given the failure of execution, led to a terrible result for the team.  Now, I focus on process more than results, but when your process is bad, you can’t be surprised by bad results.  From both an analytical and judgment perspective, Holbrook seemed to completely miss the mark.

5th inning, 0 outs, Saiko

The only successful bunt was still a questionable decision to my mind.  Saiko got the ball down and advanced Martin and English, though note English’s speed on the basepaths here was somewhat mooted by the slow-footed Martin (a deep fly ball is less helpful with Martin out there needing to get home).  And in fact, the bunt would’ve ultimately failed to plate Carolina a run but for a dropped fly ball by the Tar Heel’s outfielder.  Holbrook gave up an out with his 3rd (and best) hitter in the 1st inning, and I didn’t find it much more defensible to do the same with the leadoff man here.

6th inning, 0 outs, Bright

In the 6th, the Gamecocks ran into the same problem they faced in the first inning.  The Tar Heels, clearly aware of how up-tight Holbrook was managing the game, put the wheel play on with the goal of getting the runner at third base (in the wheel, the first baseman charges hard with the hope of throwing to third, where the third baseman remains to cover the bag – the pitcher covers the 3rd base side bunts, and the middle infield rotates toward first base).

As Holbrook said, and I agree, Bright put down a bad bunt here.  But I don’t think it’s as cut and dry as Holbrook makes it out to be.  He says a bunt to the third base side is the most successful here, but if the pitcher comes up clean, he has a short throw to the third baseman who is holding the bag, which could easily lead to the same result.  It’s asking a lot of Bright to precisely bunt right down the line in a way that the pitcher can’t field the ball so quickly as to make that short throw ahead of the arrival of the not exactly fleet of foot Grayson Greiner.

“Scoring First”

One other thing I want to note here without doing too much analysis.  The argument for Pankake bunting was to score first.  Yet, if you scroll back up to the first chart, you’ll notice the Gamecocks had a 73% chance of scoring with runners on 1st and 2nd and 0 out.  After a successful bunt, Carolina could’ve expected a 71% chance of scoring with runners on 2nd and 3rd and 1 out.  Holbrook’s decision didn’t even optimize the chances of achieving his (misguided) goal.

Overall

I think Chad Holbrook has done a great job in his first year at Carolina, and think the team will ultimately succeed under his stewardship.  To take this group to the Super Regionals and to the cusp of beating the top-seed Tar Heels was quite an accomplishment, and he nor the team have any reason to hang their heads.

That said, these were strategic blunders that harmed Carolina’s chances of winning the game.  More bothersome, these are the types of mistakes that, if you took the time to research and understand, you can avoid.  Holbrook has forgotten more baseball than I’ll ever know, but when it comes to in-game strategy, he needs to improve his own abilities to ensure that if Carolina is playing a tight game in June next year, he’s making the kinds of decisions that give the Gamecocks the best chance to win.

 

About these ads

About marvinnedick

Blogging from the mid-Atlantic on Gamecock sports, as well as general musings on sports theory otherwise.
This entry was posted in Uncategorized and tagged , , . Bookmark the permalink.

5 Responses to Bunting is normally a bad idea

  1. This is the correct analysis, and well done. I would add two things. Coach Holbrook is hardly alone in his bunting philosophy. Many managers and other baseball guys would have made similar (incorrect) decisions. It is almost the orthodox position.
    Also, I think the same analysis can be applied to the decision to steal a base. At least early in the season, we were doing that way too often as well. http://www.leftoverhotdog.com/articles/other-stuff/are-we-stealing-too-often.html

    • marvinnedick says:

      Nice read – I’d missed that earlier in the season. And it’s completely right – given the run environment, almost no college team should steal bases with less than two outs unless you’ve got a sure thing on the basepaths.

      This weekend was actually a decent time to run, given that Chapel Hill started a very weak catcher behind the plate (even on the pitch out, I thought our runner beat the throw). But overall, yes, this is another area where our coaching staff (which, as you rightly note, is not alone in this) loses an opportunity to improve our chances to win.

      I think the frustration for myself and others boils down to simply this – I can figure this out and I’m not paid to do it. Surely, given the number of coaches we have on the staff, someone could read up on this and use it to improve our decision making process?

  2. Walter says:

    Hard to argue against this analysis about bunting and base stealing. I share the frustration of your unfortunate conclusion, Marvin. Unfortunately, many people are educated in the conventional ways of doing things and apply those idle minded practices into play despite overwhelming evidence of the fallacy of their thinking. The numbers don’t lie. Over time the statistics will bare out that the inescapable truth that some of Coach Holbrook’s decisions are hurting the team. He needs to open up his mind and I believe he will. In close games against top notch competition, it’s those crucial decisions which matter the most. It seems to me that Tanner was more trusting of the bats to move the runners and more reluctant to squander an opportunity with runners in scoring position. What do you guys think about that?

    • marvinnedick says:

      If I remember correctly, Tanner had a bunt fetish early on in his career that he slowly weaned off as the years went on. Unfortunately box scores don’t go back that far so I can’t find a game that will show the point, but it frustrated me with him at times as well. Wish I had more for you.

      Of note, this isn’t just a Holbrook thing. Indiana similarly squared their number 3 hitter with two on and no one out in the first inning of their CWS game against Louisville the other night. Though to their credit, the manager improved that process in the 3rd, when in the same situation, he let the guy hit away (there’s a question as to why he needed the two innings to learn that lesson). The point is, mismanagement at the college level is certainly not a problem limited to USC. The issue is, that given all the good things a guy like Holbrook brings, it’s hard to find a way to discipline him on this issue.

      The best recent example I can think of is Toman as a third base coach. He was worth keeping around for so many reasons, but man could Toman run us out of an inning from the coaches box. I cringe just thinking about it.

  3. Walter says:

    I don’t really follow baseball too much, but I think Bobby Cox might be an example of a manager who properly hedged his bets. He wasn’t known for wreckless gambling with men on base, was he?

Leave a Reply

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

WordPress.com Logo

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

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s