All That Ned Yost Bunting Has Helped the Royals

Relative to the rest of the league, Ned Yost‘s bunting isn’t exactly out of control. However, he does seem rather fond of the strategy, so he pulls it out pretty often, and it’s a big part of how he’s labeled online. The Royals bunted and ran like crazy people in the wild-card playoff against Oakland, and in Wednesday’s Game 4 against the Orioles, Lorenzo Cain sac bunted in the first inning, with nobody out, as the third hitter in the Kansas City lineup, facing Miguel Gonzalez. The Royals did score twice in the inning, but it was taken to be another bit of good Royals luck, and the bunt predictably drew its critics. If nothing else, it looked weird. Cain, again, was batted third, by his own manager.

But there’s a funny thing about Ned Yost’s sacrifice bunts. This goes beyond just the wild-card playoff bunts mostly being defensible. In theory, a sacrifice bunt is either successful or unsuccessful. Even if successful, it trades an out for a base or two. But bunts, as you know, have a whole range of potential outcomes. The Giants, just Tuesday, won on a walk-off sac bunt attempt. 2014 Ned Yost has called for a bunch of sacrifice bunt attempts, and overall, they’ve actually been good for the team.

This is all about Win Probability Added, or WPA. When you start learning about baseball analysis you’re exposed to run-expectancy matrices, and those provide the simplest possible argument against the sac bunt, but that doesn’t tell you enough. Teams don’t play baseball games trying to get runs. They’re trying to win. Because winning is always the goal, WPA should be the preferred metric, and while win expectancy and run expectancy will be very strongly linked, they diverge in precisely the sorts of situations where a sacrifice bunt might be reasonable. Also, there’s just more to the story. Because of the other possible bunt outcomes. I’m tripping over my own words, so let’s proceed toward some numbers.

This all started because I was thinking about the Giants’ walk-off bunt from Tuesday. That got me thinking about the Nationals’ two-run bunt against the Giants earlier, and then when you’re a baseball writer on the Internet, when you think about bunts, you eventually think about Ned Yost. The Baseball-Reference Play Index is awesome, and within its searchable Event Finder, it includes sacrifice bunt attempts. So I decided to look at all such attempts on a team level from 2014, and I took the additional step of leaving out pitcher bunts because many pitcher bunts are perfectly reasonable. People generally don’t get mad when a pitcher drops one down. People get mad when Lorenzo Cain drops one down. So now look at the resulting table, of 2014 data. You see all 30 teams, and you see the cumulative sac-bunt-attempt WPA. Surprise!

Team Sac Bunt WPA
Royals 1.0
Blue Jays 0.7
Angels 0.5
Astros 0.5
Athletics 0.5
Nationals 0.5
Giants 0.4
Mariners 0.3
Brewers 0.2
Tigers 0.2
Mets 0.1
Yankees 0.1
Diamondbacks 0.0
Indians 0.0
Marlins 0.0
Orioles 0.0
Pirates 0.0
White Sox 0.0
Red Sox -0.1
Dodgers -0.2
Rockies -0.2
Twins -0.2
Phillies -0.3
Rays -0.3
Cardinals -0.4
Rangers -0.5
Reds -0.5
Braves -0.6
Padres -0.6
Cubs -0.9

In first place, having derived the greatest benefit, are the Royals. Ned Yost’s Royals, with a slim lead over the Blue Jays. This doesn’t include the playoffs, but the playoffs also wouldn’t knock the Royals out of first if considered. During the year, Royals non-pitchers attempted 52 sac bunts, as interpreted by Baseball-Reference, and they were a net positive for the team.

This doesn’t capture everything. It doesn’t capture plate appearances that featured a bunt attempt, but wrapped up in another way. It doesn’t capture the odd bunt-for-a-hit attempt with men on. But this does pretty well, and certainly it doesn’t serve as evidence that the bunting was bad for Yost and Kansas City. How did this happen? Well, lots of times, the Royals players dropped down ordinary sacrifices. But one time, Alcides Escobar reached on an error. Jarrod Dyson reached on an error. Escobar also reached on two other errors. There were a total of 13 balls in play ruled singles. It’s so easy to forget that a sac bunt attempt can result in a baserunner, since the whole point is to give yourself up, but it happens, and it can happen pretty often.

Maybe you’d argue that this is Yost getting lucky, that no one plans on the sac bunter reaching. But because it happens, it’s part of the calculation, and if sac bunters never did reach, they wouldn’t be so psychologically appealing. I don’t know how much Yost would bunt if the outs were more assured. I just know what’s actually happened, and this year, those sacrifice bunts have been more good than bad for Kansas City, even ignoring any kind of matchup analysis.

Really, this isn’t just about Yost. Although I’ll mention that, last year, the Royals’ sac bunt attempts also resulted in a positive total WPA. If you look at all of Baseball-Reference’s 2014 sacrifice bunt attempts, you see a cumulative -18.7 WPA. Very bad strategy! But now look what happens when you leave out pitchers: you see a cumulative 0.2 WPA. Essentially, the league broke even, with enough bunters reaching base to offset the outs and various failures. Again, it’s not complete, because we’re missing “expected” WPA and because we don’t know which sac bunt attempts were actually sac bunt attempts, but if you figure this is in the ballpark, suddenly it doesn’t look like managers bunted poorly at all, with their position players. There are bad bunts, but there are also good bunts, and there’s not much you can say to argue with an even WPA. That means, overall, things are neither helping nor hurting.

That’s the league. For the Royals, the bunts have helped. Not in every case, but, overall. And while that doesn’t mean they’ll continue to help in the future, it means Yost hasn’t sunk anything with the bunting to date, and maybe later on one shouldn’t be so quick to be critical. Yost absolutely makes his mistakes, but so do opposing defenses. That’s one of the ways in which a sac bunt can pay off.



Print This Post



Jeff made Lookout Landing a thing, but he does not still write there about the Mariners. He does write here, sometimes about the Mariners, but usually not.



Sort by:   newest | oldest | most voted
MGL
Guest
MGL

As you say, the WPA is very, very inexact, for multiple reasons, probably the most significant is that we have NO IDEA how the batter would do if he did NOT bunt, especially with the infield at least partially expecting a bunt. The WPA numbers you see assume league average everything, which is obviously not the case.

However, the fact that KCA has the best WPA of all the teams, and by quite the margin, suggest that their bunts are not bad (as a group – we have no idea which ones may have been bad and which ones may have been particularly good).

If you bunt with good bunters and fast runners, which is what KC does, I think, there is almost no doubt that bunting can be an effective weapon. That is exactly what we suggest in the book. That KC has such a high WPA compared to the other teams, as you also point out suggests that they are getting more than their share of hits and errors which is essential to a correct bunting strategy.

Bunting just for the sake of bunting with little regard to the speed and bunting ability of the batter, is generally a bad strategy. That is likely what those teams at the bottom of your chart are doing.

There is of course lots of random fluctuation in those numbers, so they should not be taken with a whole lot of confidence.

And let’s not even start talking about game theory, where you generally need to bunt and not bunt randomly in some fixed ratio, even with your good/fast bunters.

Costanza
Guest
Costanza

Mitchell, I’d like to thank you for your contributions especially recently in the comments. You have amazing insight into the game. While I’d love for you to write more, I understand that it’s a lot more effort than dropping a comment here.

In short, I’m really happy every time I see one of your insightful comments.

MGL
Guest
MGL

Thanks. I appreciate it. Loved you on Seinfeld!

haishan
Guest
haishan

Looking at just the 2014 data, there doesn’t seem to be any correlation at the team level between WPA on non-pitcher bunts and proxies for speed (BsR or SB). Probably that data is too noisy to see any such relationship, but it might be instructive to break it down by player, and see what kind of WPA the “fastest” hitters get on sac bunts vs. the WPA of the “slowest” hitters.

Joshua Choudhury
Member
Joshua Choudhury

I agree with this comment. WPA is a really bad tool for analyzing specific decisions like this in isolation without presenting an alternative situation (something like ‘expected WPA’). And while I think FG has, in general, done a good job of acknowledging this limitation, I think it’s a more serious one than the articles would seem to imply.

The value of giving up an Alcides Escobar PA is lower, plus the chance he reaches is higher. And Aoki is exactly the type of hitter before whom a bunt makes sense–the first-to-second advancement is relatively more valuable than if conducted in front of a TTO guy.

DavidKB
Guest
DavidKB

Wait a sec, we don’t have “no idea” how the batter would do, right? That’s exactly what the WPA is. “How the batter normally does” is baked into those numbers, assuming sac bunts are sufficiently rare as not to change the RE environment. You could perhaps argue that managers are sac bunting better-than-average batters, and that would be a fair criticism. But assuming there is no significant difference in hitting skill between league average batters and league average sac-bunters, Jeff’s WPA analysis should be totally legit. If managers are choosing to sac bunt more often with their poor hitters, the WPA analysis probably *underestimates* the actual WPA of the play.

DavidKB
Guest
DavidKB

A concise version of the above: “The average WPA of a plate appearance is 0.” This has to be true, because every game has one winner and one loser.

wer
Guest
wer

but the average wpa of a plate appearence with a runner on second and no outs will be greater than 0

Jeff
Guest
Jeff

To wer: the expected benefit of a runner on second and no outs should already be included in the Win Probability *before* the plate appearance, so the average Win Probability *Added* of the plate appearance should still be 0.

DavidKB
Guest
DavidKB

Exactly. The average WPA is *always* zero. Otherwise you have the WPA wrong.

I think Jeff was doing the safe thing and over-doubting his results. But WPA is actually made for just this type of question.

Skin Blues
Member
Member
Skin Blues

The average WPA of a plate appearance with Miguel Cabrera at the plate is much different than the average WPA of a plate appearance with Alcides at the plate. Which is a big flaw of just using WPA without using some form of xWPA. Prime example is how bad the WPA is when you include pitchers; it assumes average hitting ability so WPA thinks it’s a terrible idea for pitchers to bunt so often. For instance, I’m sure if the Angels bunted quite often with Mike Trout, their WPA in thsoe situations would be positive. He’s blazing fast, after all. But it would most certainly have a huge negative effect on run scoring by taking the bat out of the hands of the best hitter in baseball.

It should also be noted that there’s an opportunity cost of getting great WPA from poor-hitting players. And part of that is the fact that you’re giving ABs to poor hitting players. That’s not a criticism of Yost this year, as the 2014 version of Alcides Escobar had a good year at the plate with a .307 wOBA and rarely batted higher than 8th in the order. But there are teams who probably lose more value by using a crappy player in a high batting order spot than they gain in sac-bunt-WPA by said crappy player. Take, for instance, the 2013 version of Alcides Escobar, who got a majority of his ABs in the 2 spot despite an MLB-worst .247 wOBA.

wpDiscuz