Goose Eggs: fixing the save stat and reliever usage

SumnerH

Malt Liquor Picker
Dope
SoSH Member
Jul 18, 2005
31,891
Alexandria, VA
I enjoyed this article.
So did I, breaking out into a main board discussion. The article proposes a new stat, the goose egg, which is easily calculated during the game (like a save) but correlates much better with optimal reliever usage.

Essentially you earn it for clean innings in close and late games, and can earn multiple for multi inning appearances.

A relief pitcher records a goose egg for each inning in which:

  1. It’s the seventh inning or later;
  2. At the time the pitcher faces his first batter of the inning:
    1. His team leads by no more than two runs, or
    2. The score is tied, or
    3. The tying run is on base or at bat
  3. No runs (earned or unearned) are charged to the pitcher in the inning and no inherited runners score while the pitcher is in the game; and
  4. The pitcher either:
    1. Records three outs (one inning pitched), or
    2. Records at least one out, and the number of outs recorded plus the number of inherited runners totals at least three.
 

Cesar Crespo

79
SoSH Member
Dec 22, 2002
21,588
If you look at the numbers of recent closers, saves correlate to goose eggs but goose eggs don't correlate to saves. I don't know how to word it. Basically if someone has 44 saves, he probably has 40 goose eggs. If someone has 40 goose eggs, he may have 4 saves or 38.
 

LogansDad

Member
SoSH Member
Nov 15, 2006
29,053
Alamogordo
If you look at the numbers of recent closers, saves correlate to goose eggs but goose eggs don't correlate to saves. I don't know how to word it. Basically if someone has 44 saves, he probably has 40 goose eggs. If someone has 40 goose eggs, he may have 4 saves or 38.
In easy to understand terms, it basically comes down to "holds" and "saves" being equally important. Which i think is a good step in the right direction in terms of evaluating relief pitchers.
 

nvalvo

Member
SoSH Member
Jul 16, 2005
21,479
Rogers Park
In easy to understand terms, it basically comes down to "holds" and "saves" being equally important. Which i think is a good step in the right direction in terms of evaluating relief pitchers.
With one further wrinkle: you can earn more than one goose egg per game, but only one hold.
 

Savin Hillbilly

loves the secret sauce
SoSH Member
Jul 10, 2007
18,783
The wrong side of the bridge....
With one further wrinkle: you can earn more than one goose egg per game, but only one hold.
Also, you don't have to pitch a scoreless inning to get a hold; you just have to not blow the lead*. A pitcher could come into the 7th inning with a 3-0 lead and the bases empty, throw five straight walks before K'ing the next three batters, leave with a 3-2 lead, and get a hold. But that wouldn't be a goose egg.

This makes an interesting rival to Shutdowns/Meltdowns as a saves alternative. The goose egg probably has a better chance to catch on because it's easy to tell in real time, without access to a spreadsheet, whether a pitcher has earned one. But S/M is more elegant and simple in its own geeky way.

*At least, according to most if not all interpretations of the hold that I've run into; of course there's no official definition.
 

richgedman'sghost

Well-Known Member
Lifetime Member
SoSH Member
May 13, 2006
1,870
ct
Could whoever started the thread provide some context? It seems to start rather abruptly. For those not in the know, what is the difference between "Shutdowns" "Meltdowns" and "Goose Eggs"? From the article, it seems "Goose Eggs" is easy to understand.
One minor point is I don't like when mods start new threads without providing the context of why the thread needed to be started in the first place.
 

drbretto

Member
SoSH Member
Apr 10, 2009
12,074
Concord, NH
Could whoever started the thread provide some context? It seems to start rather abruptly. For those not in the know, what is the difference between "Shutdowns" "Meltdowns" and "Goose Eggs"? From the article, it seems "Goose Eggs" is easy to understand.
One minor point is I don't like when mods start new threads without providing the context of why the thread needed to be started in the first place.

The thread is split out if a different thread. It doesn't have a real introduction because it didn't start as it's own thread. This happens here pretty often and it's something they said they'd being doing more often as well.
 

DJnVa

Dorito Dawg
SoSH Member
Dec 16, 2010
53,840
In easy to understand terms, it basically comes down to "holds" and "saves" being equally important. Which i think is a good step in the right direction in terms of evaluating relief pitchers.
Are they?

I would surmise that teams that blow a save have a lower winning percentage than teams that blow holds.

Could be completely wrong on that.
 

Rovin Romine

Johnny Rico
Lifetime Member
SoSH Member
Jul 14, 2005
23,679
Miami (oh, Miami!)
Could whoever started the thread provide some context? It seems to start rather abruptly. For those not in the know, what is the difference between "Shutdowns" "Meltdowns" and "Goose Eggs"? From the article, it seems "Goose Eggs" is easy to understand.
One minor point is I don't like when mods start new threads without providing the context of why the thread needed to be started in the first place.
Let me start by saying that I really like this idea.

It's certainly a step in the right direction, but I think to measure effectiveness, you need to compare it to either overall "goose egg opportunities" and/or "blowing a goose egg" or whatever you want to call it.

The stat could be further tweaked - 7th inning RS lead by two. Reliever A inherits a runner on third, gives up an unearned run on an OF error, then strikes out the heart of the order. No goose egg. In the 8th, Reliever A strikes out two, and is lifted for a LOOGY, who strikes out the last man. Again no goose egg. But every fan takes that performance in a heartbeat.
 

BaseballJones

ivanvamp
SoSH Member
Oct 1, 2015
24,375
Good stuff in that article. But it seems to only account for whole innings = goose eggs. It then explains that two scoreless innings in the "goose egg situation" is worth twice as much as one scoreless inning. That's true. But you only get a goose egg for a whole scoreless inning. What, though, if you pitch 1 2/3 scoreless innings? That's worth 67% more than a single goose egg, and you get no credit for those two extra outs.

If that happens a lot, you could end up with a guy who pitches dozens of more scoreless innings in "goose egg situations", but gets no credit for it.

Consider this brief four-game example:

Game 1: 1.1 scoreless inning = 1 goose egg
Game 2: 1.2 scoreless innings = 1 goose egg
Game 3: 1.1 scoreless inning = 1 goose egg
Game 4: 1.2 scoreless inning = 1 goose egg

So by this formula, he gets 4 goose eggs. But in reality, he pitches 6 scoreless innings in goose egg situations, so he should have 6 goose eggs. But he only gets credit for the 4. But he's 50% more valuable than another reliever who gives them 4 scoreless innings (1.0 in each game) in those same exact situations, yet they both end up with 4 goose eggs.

So I love this concept, but it might need a little refining.
 

uk_sox_fan

Member
SoSH Member
Nov 11, 2006
1,273
London, England
I think every scoreless inning (7th+ in tight games - and I would include those where a team is down 1, even, up 1 or up 2 to be more consistent with the leverage matrix) should get goose egg awarded. If the inning's outs are split between pitchers then it's awarded to the pitcher who got 2 outs and if 3 different pitchers each got an out then the final out gets the egg. This will keep a team's total goose eggs consistent with its overall bullpen performance and thus make team bullpens easier to compare (i.e. it wouldn't penalise a team for using specialists over one that doesn't and it would minimise a manager's incentive to leave a guy in for the stat rather than what's best for run-prevention).

I don't think eggs should be awarded for innings with only unearned runs.
 

Cesar Crespo

79
SoSH Member
Dec 22, 2002
21,588
It would probably never be a problem with the way baseball works, but if a MR entered a 6-2 game in the 7th and gave up 2 runs, then pitched a scoreless 8th and 9th... he created his 2 goose eggs.
 

ALiveH

Member
SoSH Member
Apr 23, 2010
1,104
It's a little more advanced (closer to a WAR stat), but if I were a baseball exec I would have my stats guys create a stat that multiplies the leverage in the situation (which is already available) by the number of scoreless innings pitched in that situation less a constant >1 (since a run allowed is worse than a scoreless inning is good) times runs allowed times the leverage of the situation.

That being said the goose is a yuuuge improvement over saves and holds. I wish they would provide some context for what ratio of goose / broken eggs is good, bad or elite.
 

Red(s)HawksFan

Member
SoSH Member
Jan 23, 2009
20,676
Maine
Are they?

I would surmise that teams that blow a save have a lower winning percentage than teams that blow holds.

Could be completely wrong on that.
Just more evidence of how inexact the hold stat is...there is no "blown hold" stat. Blown holds are recorded as blown saves.
 

doc

Member
SoSH Member
Jul 14, 2005
4,475
Why don't you call a save anytime a reliever comes into a game with a 2 run lead or less and the tying run come to the plate you get a save when you record 3 outs or get out of the inning. A hold would be when the game is tied when you come in and when you leave. You can have multiple saves per game and I think it would better reflect how a bullpen should be used.
 

tims4wins

PN23's replacement
SoSH Member
Jul 15, 2005
37,057
Hingham, MA
I think the no inherited runners scoring piece of it is a bit harsh. If a reliever comes into a bases-loaded, no out situation, with his team up by 2 or 3 runs, then goes sac fly-double play to get out of the inning with 1 run across and his team still in the lead, that is a DAMN good performance. I think how the inherited runner scores is important. If a reliever enters with a man on 1st and no outs and that runner scores, then it is probably that reliever's "fault". But not in my bases loaded situation
 

trekfan55

Well-Known Member
Lifetime Member
SoSH Member
Oct 29, 2004
11,586
Panama
I think the no inherited runners scoring piece of it is a bit harsh. If a reliever comes into a bases-loaded, no out situation, with his team up by 2 or 3 runs, then goes sac fly-double play to get out of the inning with 1 run across and his team still in the lead, that is a DAMN good performance. I think how the inherited runner scores is important. If a reliever enters with a man on 1st and no outs and that runner scores, then it is probably that reliever's "fault". But not in my bases loaded situation
Agreed 100%. Take Game 5 of the 2004 ALCS, To Gordon comes in the 8th inning, 2 run lead. HR by Papi, 1 run lead. Then a walk by Millar, Roberts comes in, gets in Gordon's head, Nixon singles, men on 1st and 3rd, no outs and in comes Rivera. Varitek hits a sac fly and it's a tie game. Mariano gets a blown save, and negative score on this new graphic. It's not really fair to pin it all on him. And he did make it out of the inning without further damage.

Now, Game 4 is a different story. Mariano came into a clean 9th inning with a 1 run lead and blew it.
 

pokey_reese

Member
SoSH Member
Jun 25, 2008
16,247
Boston, MA
I think the no inherited runners scoring piece of it is a bit harsh. If a reliever comes into a bases-loaded, no out situation, with his team up by 2 or 3 runs, then goes sac fly-double play to get out of the inning with 1 run across and his team still in the lead, that is a DAMN good performance. I think how the inherited runner scores is important. If a reliever enters with a man on 1st and no outs and that runner scores, then it is probably that reliever's "fault". But not in my bases loaded situation
I think you are right about trying to get context in there, and that's already happening with the whole leverage part that is so central to the goose egg calculation. The most obvious solution is to weight the calculation of this perfect stat further based on game state, but now you are basically just recreating reliever WPA. It's not a bad thing, but it defies the whole 'easily calculated by a casual fan in real time.'

I guess what I'm saying is, no team should ever use 'goose eggs' in their analytics department, but if it gives fans/announcers something to talk about that isn't saves for comparing relievers, then it could still be valuable.
 

Savin Hillbilly

loves the secret sauce
SoSH Member
Jul 10, 2007
18,783
The wrong side of the bridge....
Could whoever started the thread provide some context? It seems to start rather abruptly. For those not in the know, what is the difference between "Shutdowns" "Meltdowns" and "Goose Eggs"? From the article, it seems "Goose Eggs" is easy to understand.
Shutdowns/Meltdowns is a metric featured at (and, I think, invented by?) the folks at Fangraphs. It's based on Win Probability Added and is very simple: if a reliever increases win probability by 6% or more, he gets a Shutdown. If he decreases it by 6% or more, he gets a Meltdown. If his result is in between those two points, he gets neither. The 6% figure was chosen because it produces totals that correlate well to save totals--i.e., shutdown leaders will typically be in the 40+ range.

I think the no inherited runners scoring piece of it is a bit harsh. If a reliever comes into a bases-loaded, no out situation, with his team up by 2 or 3 runs, then goes sac fly-double play to get out of the inning with 1 run across and his team still in the lead, that is a DAMN good performance. I think how the inherited runner scores is important. If a reliever enters with a man on 1st and no outs and that runner scores, then it is probably that reliever's "fault". But not in my bases loaded situation
This is the attraction of the Shutdown/Meltdown stat. It correctly sees the value of that bases-loaded, no-out, one-run performance as roughly equivalent to the value of a one-out, 2nd-and-3rd, no-run performance, while the Goose Egg draws an arbitrary line between the two. (The S/M stat has an arbitrary line of its own, the 6 percent thing--a pitcher who increases win probability by 6.0 percent is considered to have done the same thing as a guy who increases it by 9 percent, but a much better thing than a guy who increases it by 5.9 percent, which of course makes no sense.)
 

SumnerH

Malt Liquor Picker
Dope
SoSH Member
Jul 18, 2005
31,891
Alexandria, VA
Could whoever started the thread provide some context? It seems to start rather abruptly. For those not in the know, what is the difference between "Shutdowns" "Meltdowns" and "Goose Eggs"? From the article, it seems "Goose Eggs" is easy to understand.
One minor point is I don't like when mods start new threads without providing the context of why the thread needed to be started in the first place.
I split it out from the game thread, and tried to provide a brief summary in my post (which is the 3rd post of the thread).

Shutdowns and meltdowns are unrelated to the article--they're a Fangraphs stat Savin Hillbilly brought up. They're nice, but they require calculating WPA which is difficult to do by hand in real time.

Goose eggs are designed to be easy enough for managers to figure out in their head during the game, but much better than saves for driving reliever usage.

Let me start by saying that I really like this idea.

It's certainly a step in the right direction, but I think to measure effectiveness, you need to compare it to either overall "goose egg opportunities" and/or "blowing a goose egg" or whatever you want to call it.
The article talks about that in a fair amount of detail, and comes up with a goose-egg WAR and some other more sophisticated stats based on the relative value and occurrence of goose eggs vs. broken eggs and the like.
 

SumnerH

Malt Liquor Picker
Dope
SoSH Member
Jul 18, 2005
31,891
Alexandria, VA
...It's not a bad thing, but it defies the whole 'easily calculated by a casual fan in real time.'

I guess what I'm saying is, no team should ever use 'goose eggs' in their analytics department, but if it gives fans/announcers something to talk about that isn't saves for comparing relievers, then it could still be valuable.
Not just fans, but managers. You're not allowed to have computers in the dugout--analytics departments are great between games, but they can't feed you in-game info. Part of what's driving reliever usage now (IME, and the article seems to agree) is that managers want to get saves for their closers and can easily figure out in-game when to deploy people to get saves.

If you can make people appreciate goose eggs more than saves so that managers decide to go for goose eggs instead, that'll help push reliever usage back in a more optimal direction.

Ideally managers would be looking at a football-style wristband with a precalculated grid of game situations for reliever usage, but that's not happening for whatever reason. If they're going to be slaves to stat accumulation, at least give them a better stat to target.
 

Toe Nash

Member
SoSH Member
Jul 28, 2005
5,595
02130
If we want to get from where we are now to "optimal reliever usage", isn't the problem
1. that saves are overvalued by the marketplace (due to the idea that a "closer" is somehow more special than any other kind of reliever), leading to players caring a lot about getting them, and
2. that players like to have a defined role, and closer is a defined role built from the save stat

I don't see how developing a new stat solves either of these without creating new ones. Problem 1 is lessening as analytics use increases by front offices, but having a new stat doesn't really help -- smart front offices are already looking at things like WPA and all the other advanced stats to make decisions about how much to offer relievers. Dumb front offices maybe aren't, but they can't be very numerous nowadays and giving them another flawed statistic (even if it's better) is only kind of solving the problem.

Problem 2 is just going to continue as guys will want to be in a position to get "goose eggs". This is maybe a little better than holds or saves but they'll still want to be used in the 7th or later, when maybe the 6th is actually the most important spot in the game.

What needs to happen is to dissolve all connection between saves and compensation. A good step toward this was in reforming the free agent compensation system from a stat-based one (using bad stats to determine the level of compensation) to a market-based one (allowing teams to determine if they're giving the player a qualifying offer). This is still vulnerable to FOs overvaluing saves, but at least it's not automatic anymore.

If WPA is the best stat for relievers, we should just use WPA (or WPA with a combination of other stats). Do many people use RBI as a measure of a hitter's value anymore? Did we need to get Runs Created into major usage in order for that to happen, or did it just slowly die? The same thing will happen with saves, sooner or later, because they are a flawed stat.

As for in-dugout use by managers, they may not be able to look at the live WPA graph, but any good manager should know intuitively when a high-leverage situation happens. Right now they're saving their closer for the 9th, but a good part of that is because the closer likes having a defined role and because saves make them money - not because they can't figure out when a better time to use them might be.
 

SoxJox

Member
SoSH Member
Dec 22, 2003
7,082
Rock > SoxJox < Hard Place
Are we talking about Anser, Branta, or Chen geese?

And, as far as
goose eggs vs. broken eggs and the like
is concerned, can we work scrambled eggs, omelets, and quiche into the mix? All this talk of eggs is making me hungry.

On a more serious note, I think Sumner touches on an excellent point re. being designed to be easy enough for managers to figure out in their head during the game and the potential to lead to better reliever usage. It really could lead to many more exciting instances of situational decision making and climactic points within a game.
 
Last edited:

Wake's knuckle

New Member
Nov 15, 2006
565
Aarhus, Denmark
I'm not sure I necessarily like the way it deals with runner on base when entering the game. Getting out of a bases-loaded, no out situation without allowing a run is much much more valuable than coming in with the bases empty.
 

AB in DC

OG Football Writing
SoSH Member
Jul 10, 2002
13,620
Springfield, VA
I think people may be setting the bar a bit high here. I don't think 538 proposed this as the be-all and end-all of reliever counting stats. It's a useful measure, though, and fairly easy to think about in real time. That's a good thing. Further refinement might be better, but there's something to be said for simplicity. I think this hits the right balance very nicely.

[Or to put it another way, it's more of a "fun" stat than a sabermetric one.]