clock menu more-arrow no yes mobile

Filed under:

Bunting and the new extra-innings rule

Divisional Series - St Louis Cardinals v Atlanta Braves - Game Five Photo by Kevin C. Cox/Getty Images

MLB officially rolled out a series of rule changes for the 2020 season yesterday. Among them are a (definitely won’t be enforced) ban on spitting, use of the DH in National League games, and an extra-innings procedure similar to that in Minor League Baseball. As the title of this post suggests, we’ll be honing in on extra-innings today. To summarize the new rule:

  • Every half-inning in extras will begin with a runner on second base.
  • The runner on second will be the player in the batting order directly before that inning’s leadoff hitter, unless the manager elects to make a substitution. This would function like any other pinch running appearance, so the removed runner can’t return to the game.
  • If the pitcher is somehow slated to be the runner—virtually impossible given the implementation of a universal DH—the batting team may place the player in the order before the pitcher on second instead.
  • If you’re worried about this rule affecting your fantasy team, know that the initial runner on second will be considered an unearned run if he comes around to score. The scorekeeper will treat that runner similarly to a runner who reached second on a leadoff error.

In a way, this extra-innings setup mirrors overtime in college football. In the latter, both teams start with identical field position at the 25-yard line, but there is an advantage to be had by going second. If you’re on offense second, you know what you need to score to either extend the game or win. If your defense held the opposition to a field goal or no points, you have the luxury of going into your drive knowing you can settle for a field goal. If they scored a touchdown, you know you have to match them, allowing you to make more informed play calls, especially on fourth down.

Likewise, the team batting in the bottom of the 10th has a clearer picture regarding their win conditions. This has always been the case in baseball to some extent, but if the other side came up empty or only scored one run in their half of the inning and now you automatically start with a runner in scoring position, it makes intuitive sense to try to “manufacture” a single run.

One of the most frequent ways teams try to play “small ball” is with the good ole sacrifice bunt. Move the runner over to third, follow it up with a deep fly ball, game over. The idea sounds good on paper. Assuming games do occur furiously knocks on wood, we’ll surely see extra-inning bunts deployed from time-to-time, both by road teams simply looking to give their closer the ball with a lead and by home teams that have no need for a big inning.

It shouldn’t come as much of a surprise that, upon my indoctrination into the cult of sabermetrics however many years ago, I found myself a member of the anti-bunting camp. A sizable portion of VEB readers probably agree with me that bunting is a generally counterproductive strategy. At this point, chances are you buy the math or you don’t. Either way, let’s run through the relevant numbers behind the question: to bunt, or not to bunt?

146 bunts were put into play (more on that in a little bit) with a man on second and nobody out in 2019. Here are the results from those bunts.

Bunt results, runner on second and no outs

Outcome Amount Share
Outcome Amount Share
Sacrifice bunt 102 69.9%
Single 19 13.0%
Batter out, no runner advance 18 12.3%
Fielder's choice, out 6 4.1%
Fielder's choice, no out 1 0.7%

Right away, it’s worth noting that roughly one in every six bunts (16.4%) ended with an out and the runner failing to safely reach third. Bunting against big league pitchers is still extremely difficult, and by no means a guaranteed success.

More important, though, is how these results translate into base-out situations. At any given moment, an inning is in one of 24 possible base-out states ranging from bases empty with no outs to bases loaded with two outs. From 2010 to 2015, the situation denoted -2-, 0 outs (i.e. the one all extra-innings will begin with) produced a run expectancy of 1.100, meaning the batting team scored, on average, an additional 1.100 runs between that point and the end of the inning.

Run expectancy is a useful measure to tell us whether one base-out situation is more desirable than another, but it’s not without limitations. For example, there might be scenarios in which we care more about scoring one run than as many as possible. With a runner on second and no outs, teams score at least once 61.4% of the time.

So those are our baseline numbers that every extra-inning begins with: a 1.100 run expectancy and 61.4% probability of scoring at least one run. What we want to find out is whether or not bunting typically improves those figures.

Base-out results after bunting

Base-out situation after bunt Run expectancy Probability of at least one run Count Share
Base-out situation after bunt Run expectancy Probability of at least one run Count Share
Total 1.051 65.4% 146 100.0%
--3, 1 out 0.950 66.0% 95 65.1%
1-3, 0 outs 1.784 86.0% 21 14.4%
-2-, 1 out 0.664 39.7% 19 13.0%
1--, 1 out 0.509 26.5% 5 3.4%
-2-, 0 outs (+1 run) 2.100 100.0% 2 1.4%
---, 1 out (+1 run) 1.254 100.0% 2 1.4%
-23, 0 outs 1.964 85.2% 1 0.7%
1--, 0 outs (+1 run) 1.859 100.0% 1 0.7%

The average run expectancy after a bunt is lower, but the odds of getting at least one run home slightly increase. So, case closed? Bunting hurts your chances at a big inning but is worth it if you’re playing for a single run? Not so fast. Remember that this data only looks at the bunt attempts that were successfully put into play in fair territory. As it turns out, only about 49-50% of bunts actually wind up fair; the rest are either fouled off or missed entirely. A foul bunt with two strikes has the same effect as a popup bunt that is caught, tanking your odds of getting on the scoreboard from 61.4% to 39.7%.

If we assume approximately 12.5% (.5 raised to the third power) of batters will strike out following three failed bunt attempts, the probability of scoring after calling for a bunt dips to 62.2%. Granted, this rough math doesn’t account for batters being able to pull their bat back at the last second against a tough pitch. It also doesn’t consider that many hitters will ditch the bunt after picking up two strikes, but, in that case, you’re still in a worse spot than had you just swung away from the start of the at-bat.

Furthermore, I believe the post-bunt numbers are artificially inflated by bunt singles that are less likely to occur in a tight extra-innings game. Of the 27 bunts in our sample that resulted in no outs being recorded and/or a run scoring, the infield was only playing in a “standard alignment” nine times. Most of the infields hits propping up the aforementioned numbers came from predominantly left-handed batters facing some semblance of a shift. Sure enough, the defensive team was less likely to give the hitter a shift when the runner at second represented the tying or go-ahead run. In other words, the statistical upside, which was the reason bunting initially looked appealing, is limited in extra-innings because the dream scenario (a bunt single that results in everybody being safe) becomes more difficult to pull off.

We can also take a look at how bunting affects win expectancy in an extra-innings situation. Using the gregstoll win expectancy tool, I converted the base-out states from the above table into the probability a home team would win if they were tied or trailing by one, the two most likely spots a team would bunt in during extra-innings. I would have liked to make similar calculations for away teams, but win expectancy data compiled from past games wouldn’t account for the home team also starting with a runner on second and no outs in the bottom half of the inning. (Note: I used data from 2000-2019 and combined all innings from the bottom of the ninth onward to procure a larger sample size.)

For context, a home team starting the bottom of the 10th (or any subsequent inning) under the new extra-inning rules has an 80.7% win expectancy in a tie game; 44.5% if trailing by one. Here is how bunting alters those numbers.

Win expectancy after bunting

Base-out situation after bunt Win expectancy (tied) Win expectancy (down by one) Count Share
Base-out situation after bunt Win expectancy (tied) Win expectancy (down by one) Count Share
Total 81.3% 43.2% 146 100.0%
--3, 1 out 80.5% 42.5% 95 65.1%
1-3, 0 outs 91.0% 59.4% 21 14.4%
-2-, 1 out 73.4% 26.7% 19 13.0%
1--, 1 out 65.9% 18.7% 5 3.4%
-2-, 0 outs (+1 run) 100.0% 80.7% 2 1.4%
---, 1 out (+1 run) 100.0% 60.6% 2 1.4%
-23, 0 outs 93.2% 65.3% 1 0.7%
1--, 0 outs (+1 run) 100.0% 73.1% 1 0.7%

The same caveats I made earlier apply here as well. Even before making adjustments to the numbers, bunting decreases your win expectancy if you’re entering the frame down by one. In a tie game, bunting ostensibly helps your chances by 0.6%, but that change in win expectancy quickly flips to a negative number once you bake in the possibility of foul bunts. If you lower the odds of a bunt hit in a close game, that negative number grows even larger.

The ramifications of bunting might be overstated given how much ire teams can draw. That said, a dip in win expectancy is a dip in win expectancy, which, after enough time, will eventually catch up to you. Each individual game suddenly becomes more important in a shortened season. So too does every decision within those games.