We’ve seen three baseball games so far this year, and if that’s a meaningful sample, I think Hyun-Jin Ryu has all the major awards locked up.
We’ve also seen eight plate appearances this season that ended with bunts. After the jump, a few facts about those bunts, which will accomplish nothing except establishing a baseline for a feature I hope to bring back to these pages with some regularity throughout the summer.
I’ll use the term “bunt” below to represent a plate appearance ending with a bunt. A player showing bunt and walking or bunting foul twice, only to swing away with two strikes, will not show up here, since Baseball Reference’s play-by-plays only note bunts that result in an out or a player reaching base (or, often, both). Of the eight bunts we’ve seen in 2014:
5 (62.5%) were bunted by pitchers. The other three were all by shortstops- Dee Gordon (twice) and Everth Cabrera.
1 (12.5%) was clearly laid down with a base hit in mind. Gordon bunted to lead off the sixth inning in the second game in Australia last week. The other seven were in sacrifice situations, so even if the bunter hoped to reach base, the primary motive seemed to be advancing the runner(s).
3 of 7 (42.9%) sacrifice attempts successfully advanced the runners(s). Others ended in a force out, a foul out, and two strikeouts.
2 of 8 (25%) total bunts put the bunter on first base- Gordon’s hit and Cabrera’s eighth inning sacrifice Sunday night, which forced Brian Wilson’s error, leaving two on with no outs and igniting the game-winning rally.
1 of 4 (25%) of runners advanced via sacrifice scored- Yasmani Grandal, who moved to second on Gordon’s error-inducer, stole third, and scored the go-ahead run on Chris Denorfia’s single.
2 of 8 (25%) total bunts led to an increase in win expectancy for the batting team. This includes the single and the reached-on-error, but neither of the successful sacrifices dropped by Ryu and Andrew Cashner.
Average change in win expectancy as a result of a bunt has been negative 0.88%, as Wilson’s error was the difference in Sunday night’s game, swinging win expectancy by 10% after all other bunts had averaged a drop in win expectancy of almost 3%.
I’ll try to keep up with these numbers to the best of my ability as the season goes on. What I’d like to hear from you, dear reader, is how you think the percentages above will compare to year-end trends. Obviously, very few of the bunts attempted by American League teams will come from pitchers, and the junior circuit bunts far less, so I’ll probably limit the study to National League bunts.
Will more than one in eight bunts come in non-sacrifice situations, perhaps as players try to take advantage of increasing use of defensive shifts? Will more than a quarter of bunts lead to increased win expectancy? How will that trend compare to the percentage of runners advanced by bunts who come around to score later in the inning? Will the average change in win expectancy for the bunting team be positive or negative? By how much?
You’re up. Are you laying one down or swinging away?
Today, Jeff Samardzija laid one down into a 1-5-3 DP, quashing a 2-on, no-out 5th inning. The Cubs went on to lose, 1-0 in 10 innings.
Samardzija is a .111 hitter, but has 19 sacs in 22 tries, and 7-for-7 with 2 aboard. So I can’t knock that strategy. But I think the majority of sac bunt tries will produce negative WPA.
If a successful sac bunt by your every day .100/.125/.150 slashing NL pitcher is -3% WPA, what’s them flailing away and striking out on three pitches worth? I’m guessing WPA doesn’t consider the expected outcome of letting your pitcher swing away, but I’ve got to think a sac bunt by the pitcher isn’t a terrible thing.
Now I’m not advocating bunting on a regular basis but it’s got to be more valuable to bunt this year than it was in, say, 1997, and there must be somewhat less negative value in pitcher bunting than pitcher hitting, right?
The standard WPA calculators and charts are all based on some concept of an average hitter being at the plate. You can’t use those calculators as a guide to deciding whether it makes sense to have a pitcher sac bunt or not (though it will give you the WPA tradeoff, more or less, between the pitcher bunting and a pinch hitter coming in to swing away). The strategy decision between pitcher bunting and pitcher swinging, as brp correctly points out, has to be based on an evaluation of expected WPA if the pitcher swings away vs. if the pitcher bunts, which you won’t get from the established WPA calculators and charts.
Usually when sabermetrics folks talk about the general inadvisability of sac bunting, they are referring to sac bunting by non-pitchers.
Last season, there were 1,372 PAs in which a pitcher was batting with at least one man on base and less than two out. 559 of those led to sac bunts that went as planned (batter out, runner or runners advance).
Agree completely, bstar. This is why I’m interested to track what percentage of bunts come with a pitcher at the plate. If it turns out that the average bunt returns a -.7 WPA and the average pitcher PA returns -.9, maybe pitchers should actually be bunting in more situations. k
That was brp, not me.
Having a name that’s commonly misspelled, I judge people a lot for seeing Bryan in an email and typing Brian anyway. This is worse. Sorry, guys.
According to B-Ref’s Play index, in 2013 there were 2,114 sacrifice bunt attempts. These produced, among other things, 1,383 sac bunts, 155 Ks, 191 hits and 64 reached on errors. The overall net WPA for those 2,114 attempts was -20.6.
Of those 2,114 total sac bunt attempts, 863 were by pitchers, resulting in 559 sacs, 6 hits and 20 reached on errors, with a total net WPA of -18.4. That would mean that the non-pitcher sac bunt attempts led to a overall net WPA of -2.2.
In 2013, there were 813 plate appearances when pitchers did not attempt a sac bunt, despite being in a sac bunt eligible situation (at least one man on, less than two outs). In those situations, pitchers had 774 at bats, 22 walks, 7 HBPs and 10 sac flies. The overall net WPA from those 813 PAs where the pitcher could have but did not try the sac bunt was -18.9.
So in 863 sac bunt situations when pitchers tried the bunt, it resulted in a net -18.4 WPA. And in 813 sac bunt situations when the pitcher did not try the bunt it resulted in a net -18.9 WPA.
So it appears it is correct for a pitcher to bunt in a traditional sacrifice situation (as long as they can bunt reasonable well and are not one of the very few pitchers who hit *much* better than a typical pitcher).
OTOH, depending on who those other hitters are doing sacs, given how close to 0 their -WPA was, if they were very weak hitters, likely to have a -WPA anyway, perhaps those sacs were also good strategy. It may be that they were choosing sacs only in situations which were more advantageous to the sac than an average “men on base, <2 outs" situation, and it seems they were primarily either poor hitters, or players with enough speed to lose less OBA from a bunt than the average player.
It seems plausible from your stats that most of the bunts in 2013 were strategically warranted, and if anything, a few more of the pitcher ABs probably should have been bunts.
Well, birtelcom, I guess I don’t have to put a year’s worth of work into this question. I’d be fascinated to know how you pulled this. Were you able to identify both sac bunts and bunts for hits?
Also, were those 155 Ks all fouled third strikes? If the hitter bunted foul twice and swung away the third time, the play index would see a strike out swinging, right? I’m afraid failed attempts will cloud any attempt to understand the value of a sac bunt attempt more than any other factor.
@21
Bryan: In a nutshell, to find the info, go to the PI, select Event Finders, Batting by Team and select the appropriate parameters.
Thanks for running the numbers on that; I’m actually surprised it’s as close as it is, really.
It’s probably a moot point for the horrible, but likely inevitable, day when NL adopts the DH, but very relevant to today’s game, at least.
John, I’m guessing your last sentence is correct, but I wonder if it’s still true if we remove the “sac”. What I think more people around the game are realizing is that the average successful sac bunt, if we define success as giving up an out in favor of moving a runner up, results in a negative WPA. What I’m interested to learn is whether the opportunity for a hit/reached on error, which will be more productive in terms of WPA than sacrifices are counterproductive, actually encourages bunting more often, particularly given the increased use of defensive shifts.
I did a quick and dirty incomplete analysis I ran the PI for games in 2013 in which a batter had a SH in his only batting appearance. There were 120 such occasions. 15 of them resulted in a positive WPA and the other 105 resulted in a negative WPA. The best thing to do is to retrieve a WPA calculator via Google and run it for various base-out situations and vary the inning and score differential. I’ll try to do a few.
Here are results for some situations from the WPA calculator.
First situation:
Bottom of the 9th, score tied.
Runner on first, no outs: WPA = 0 .72164
After SH, runner on second, one out: WPA = 0.71665.
Net change = -0.04486
Second situation:
Bottom of the 9th, score tied.
Runner on first, one out: WPA = 0.66676
After SH, runner on second, two out: WPA = 0.60575.
Net change = – 0.06101
Third situation:
Bottom of the 9th, home team down one run
Runner on first, no outs: WPA = 0.30902
After SH, runner on second, one out: WPA = 0.27756.
Net change = -0.03146
Fourth situation:
Bottom of the 9th, home team down one run
Runner on first, one out: WPA = 0.20489
After SH, runner on second, two out: WPA = 0.12625.
Net change = -0.07864
The numbers don’t quite add up for the first scenario. If the two WPA values are correct, the net change should be -.00499, an order of magnitude smaller than the other scenarios. Interesting that SH still can’t manage to break even in such situations.
Thanks for the correction.
I should mention that there is more than one website with the WPA calculator. I used the one from the gregstoll site. Tom Tango has the calculator on one of his sites. Try tangotiger.net/welist.html.
Thanks for this, Richard. I may use these numbers in future work. As I alluded to above, I think we sell managers short when we assume sacrifice is the sole motivation for every bunt a player attempts in these situations. A lot can go wrong in fielding a bunt.
I don’t think the Diamondbacks win Game 7 in 2001 if Damian Miller doesn’t lay down a bunt in the ninth. Before the play, Arizona’s WPA was 36%. As you demonstrate above, a sacrifice would have cost the D’backs .3 points, but Rivera throwing errantly to second swung WPA by 18%, to 54%. Setting aside other scenarios, that .3-point gamble pays off if there’s a 1-in-60 chance that both runners end up safe. That overstates the case by ignoring strikeouts and double plays, but it also ignores the relative futility of swinging away against Rivera.
Interesting that they’re all negative. Though they’re averages. Could a certain type of hitter coming up in the inning make any of them positive?
eorns, see Birtelcom at #9 above. The advisability of a bunt absolutely varies by hitter, and that is not borne out in WPA.
It also varies by speed of the runner, bunting skill of the bunter, even fielding. A bunt down the first-base line is less advisable if you’ve got a young Keith Hernandez ready to pounce on it as opposed to an Adam Dunn type of fielder.
Agree completely, brp. 🙂
Not a problem, Byron.
Sacrifice bunts are but one part of the reason the 2013 Red Sox had the AL’s best offense. According to yesterday’s broadcast (I believe it was Sutcliffe but I may be mistaken), the Red Sox do so well because they have a lot of productive outs.
In fairness to him, he was including things like “making an out after working the pitcher for a few extra pitches” as productive outs. But still – we deserve better than that. I guess it’s just too obvious to say “The Red Sox scored the most runs because they had the highest OBP and SLG of any AL team.”
The Red Sox were second in the AL in 2013 in the average number of pitches per PA that their hitters saw. Twins hitters saw 4.02 pitches per PA on average, while the Red Sox saw 4.01. The league average as a whole was 3.86. The White Sox, at the bottom of the AL in this stat in 2013, saw 3.75 P/PA on average. Although they were very close in P/PA, the Twins averaged 3.79 runs scored per game while the Red Sox averaged 5.27 (league average 4.33).
why use WPA instead of RE24? I suppose we miss some late inning leverage stuff but it also makes looking at pitchers bunting when well ahead or well behind more reasonable to evaluate. I’d say when looking at non-pitchers, WPA needs to be considered but for pitchers, RE24 is probably ideal.
Your comment may be related to what I just saw – literally just saw. Alex Avila bunted Austin Jackson to 2nd base in the bottom of the 10th inning with zero outs and the score tied 1-1. Whether the Tigers score 1 run or 4 is irrelevant at this point as they just need 1. The RE24 almost certainly decreased in this situation, but the Tigers are not trying to score a ton of runs, but just want 1 for the win. I would guess that neither WPA nor RE24 is exactly what we want in all situations. Sacrifice bunts are a strategic choice used in specific instances (maybe someone accidentally lays down a sac bunt, but I doubt it), and are different than someone going up to the plate to “get a hit (of whatever type),” so I’m not sure looking at them with the standard methods (if WPA and RE24 are “standard methods” – well, they are standard for HHS) is the best approach. That being said, I have no alternative to offer.
And the Tigers did score 1. Productive outs win ballgames. Sutcliffe. QED 🙂
I’m going to try to be more complete in my thought about sacrifice bunts as a “strategic choice”. Runner on 1st, no outs, a successful sac bunt is equivalent to a ground ball to third with the runner moving that ends in a 5-3 putout at first base with the batter being retired and the runner now on second. I would guess that WPA and RE24 don’t really care “what” happened to get to the situation just that we are now at the current situation (I’m thinking of something like a Markov process where the current state is what matters for the future and the history is irrelevant for the future).
The difference between the sac bunt and that ground ball out is that batters make ground ball outs all the time – it’s just a part of going up to the plate and trying to get a hit. However, sacrifice bunts are either (1) used by specific players (particularly pitchers) or (2) in very particular circumstances (like the one the Tigers just faced). So yes, the current state ends up being the same but the sac bunt is a choice to get to that current state. Perhaps WPA takes that into consideration.
As an aside, while perusing some 2014 Topps baseball cards I noticed that the last statistic on the card back was WAR.
Pingback: High Heat Stats » A Wild Weekend of Bunts
Pingback: High Heat Stats » Bunts and RE24