OT: How to Improve Upon Baseball's Save Statistic
Nate Silver at FiveThirtyEight has developed an interesting alternative to baseball's save statistic. The link is obviously where you'll find more information, but the short version is:
- Many teams use their closer exclusively in the 9th inning, even when their talents would be better served in higher-leverage situations in the 7th or 8th innings.
- This habit developed from managers' tendencies to try to get the most possible saves for their closer, instead of the most possible wins for their teams.
- Silver developed a statistic called the "goose egg," which rewards relief pitchers for a scoreless frame tossed in the 7th inning or later.
Much more thorough analysis can be found below. I encourage anyone who loves stats to give it a look.
https://fivethirtyeight.com/features/goose-egg-new-save-stat-relief-pit…
I'm not sure if this is the right answer, but I would be in favor of a better stat. Silver is hardly the first guy to raise this issue.
I'd be interested in re-labeling backdoor wins that a relief pitcher gets for giving up a lead but being the pitcher of record when his team re-takes the lead in the next half-inning, too. I remember Jose Valverde blowing a save on a Verlander gem on opening day a few years ago and picking up the win, thinking what a travesty it was.
I agree. For as well-known as baseball is (and rightfully so) for providing us with endless advanced data, the sport includes some truly broken counting stats, like wins and saves.
April 20th, 2017 at 10:04 PM ^
The counting stat issue is a double-edged sword. Obviously, the last 15 years have been a revolutionary period for hte use of advanced stats. That revolution has spread to other sports as well, and one may well consider Mgoblog's advanced analysis in the UFRs to be an extension of this trend. For the dedicated fan it has been fascinating.
Problem: Baseball has benefited from counting stats a great deal in its rich history. More than any other sport in the world, Baseball has told its story through statistics. 61, 714, 56; all numbers that mean something, that most people (at least, my age) know the significance of without me every identifying what they count. And on a more accessible basis, counting stats still mean something to casual fans for regular players. We know that 20 wins is a good season for a pitcher, that .300 is a good batting average, that 30-40 home runs are very good totals for a slugger in a season.
We weren't a "baseball family" when I grew up, but we watched baseball and my Dad would talk about the game. I learned this stuff. If you liked sports you knew what these numbers meant. And, uniquely in sports, you could understand these numbers across different eras. Babe Ruth accrued numbers in the same categories that Ken Griffey Jr. did.
Every year pundits gripe that "Baseball isn't promoting its stars enough." They look longingly at the NBA, a sport with a completely different structure, and they're jealous of it. But Baseball does not, in my opinion, lend itself as well to today's star-heavy culture. Part of that, frankly, is the depreciation of the old counting stats.
That depreciation has come from both sides; both advanced statisticians finding better stats, and cheaters devaluing the old ones. Barry Bonds broke all of the home run records, and I couldn't even tell you what those records are anymore--because I know they were acquired by doping, and I knew that it was happening at the time. The all-time hit king has a lifetime ban from baseball. The only stat left that has any weight is Dimaggio's hit streak, but that may never be approached.
Now the biggest stars are guys like Mike Trout. But Mike Trout doesn't produce spectacular highlights--some of his acrobatic catches are nice, but they pale next to a great dunk or a spectacular touchdown. His home runs are nice, but all home runs look exactly the same on Sportscenter (or a viral twitter post). The thing that could set him apart would be his stats. His average, or his home runs, or whatever.
But the stats that define how truly great he is, the advanced stats, stats like WAR (whichever version you use) or OPS, or others, don't roll off the tongue, and they lack the gravitas and the familiar numbers that the more traditional stats do.
Yeah, they're a step up in many ways. But there's been a cost.
April 20th, 2017 at 10:19 PM ^
While an amazing stat, it doesn't roll off the tongue. Most people that follow baseball would not know that number off the top of their head like they do 61 homers for Maris, or 714 for Hank Aaron or 56 for DiMaggio.
April 20th, 2017 at 10:24 PM ^
April 20th, 2017 at 11:20 PM ^
Not for Detroit Tigers closers in the last 5-7 years.
April 20th, 2017 at 10:31 PM ^
And he does a fair job of showing how relevant it is to a reliever's effectiveness in his article. I also like the related "broken eggs" stat he proposes. True, it doesn't isolate the pitcher like FIPS or something, but I think it remains relevant both to a SABRmetrician and a casual fan looking for a countable stat.
April 20th, 2017 at 11:38 PM ^
April 21st, 2017 at 12:20 AM ^
I think you're taking my argument a bit further than I am. I don't dislike advanced stats, and I think they illuminate the game in many ways.
But that doesn't mean that everything is a positive, and there has been a cost. Your average sports fan would have to use Google to discern whether or not a certain player's OPS number is good or bad. But they know a guy batting .225 isn't cutting is and a guy batting .350 is doing great.
Doesn't make OPS inferior or the guys using it bad. I don't even mind guys re-evaluating players based on more advanced information. But it does make things less accessible.
April 20th, 2017 at 10:42 PM ^
My daughter's little league commish agrees with you wins are not important. Also, runs. Only smiles.
Sometimes I wonder if Verlander is even having fun out there with his buddies (this should be the real stat). S/9. Maybe everyone would smile more.
Deep down, isn't happiness the only statistic that matters.
April 20th, 2017 at 10:12 PM ^
Maybe those should be called "team" wins, like team rushes/passes in football and team rebounds in basketball - don't give any individual player responsibility.
Wins has only been so sore spot for me. The obvious backdoor win you mention always drives me nuts. A starter can go out there and pitch his butt for 4 2/3 and give way to a guy that finishes off the 5th by getting one out. The reliever can get the win. It really is a flawed way of awarding wins but it is for the most part very objective. I think it needs to be objective but it certainly could use an overhaul.
The other stat that always bothered me is the HBP. This goes down as nothing - no at bat and no walk. I think if you get plunked you at least deserve to be credited with a BB. A lot of these big leaguers have made an art out of it and should be statistically rewarded for it.
Statistically, a HBP is exactly the same as a BB; they just get counted in different categories. The credit for the batter is the same: no at-bat, but an uptick in OBP. They just keep separate counts of walks and HBPs, proably so that people can get more statistical credit for taking the beating.
Joe Maddon is the only Manager I have seen over the last many years to sometimes leverage his closer in the 7th or 8th depending on the siutation.
April 20th, 2017 at 10:13 PM ^
April 20th, 2017 at 10:22 PM ^
April 20th, 2017 at 10:26 PM ^
There are several baseball theories that I don't get...the big one for me is having a 'starting pitcher' start a game. Normally, a starter faces the first batter and goes 5-8 innings into the game. He faces generally the best offensive players in the first inning and a half (1-5 hitters). Those hitters will see him 3 to 4 times a game, allowing them to capitalize at a higher rate and do more damage in the later innings against a tired pitcher.
I always thought if you used relief picher to face the first 4 to 5 batters, the starter would see the back end of the order more than the front, minimizing the damage that incurs due to gained info of seeing pitches each time through the order. I feel that this would also let the starter generally go deeper into a game and make less taxing pitches...
April 20th, 2017 at 10:40 PM ^
April 20th, 2017 at 10:59 PM ^
specialized pitchers faceing the better hitters. Starters don't typically have better stuff then relievers...the have better stamina and control. Specialists can go and let loose. What is the difference from treating the first inning+ like the 7th or 8th inning? The difference is that your top specialists take on the top of the order leaving the bottom for the starting pitcher to face 4 times or so before they are pulled instead of the top of the order.
April 20th, 2017 at 11:26 PM ^
Actually, starters tend to have more pitches than relievers in addition to endurance and control. Guys that slide into relief pitching often get looks as starters first and can't cut it, but can work as relievers because they don't need to see the order multiple times and, as you said, open the throttle more. Endurance is something that is conditioned, not something that is an innate attribute like a good curveball.
At any rate, this is a bad idea. You want your starting pitcher doing his best work against the top of the order, and even if the reliever does well you're talking about using him in a situation where you don't know if it's worthwhile. Key relievers aren't put into games that aren't close. So if your team has an 8-run lead by the 5th inning, you've used up a good reliever in a blowout.
The goal isn't to "save starters." It's to win games.
Every. Single. One.
April 21st, 2017 at 11:55 AM ^
In 1972, Earnshaw Cook suggested exactly this (in an article in Sports Illustrated by Frank DeFord): https://www.si.com/vault/1972/03/06/576391/it-aint-necessarily-so-and-n…
This applies more to the National League today than the American League (or any other league), but the idea is to have the starting pitcher go 2 innings, give or take, pinch hit for him the first time through the order, and then put in the "long" pitcher, who will pitch about 5 innings and bat once. At that point, you're through the 7th, and can decide the best course based on the score.
He suggests it could score an additional 50 runs a season with the extra use of pinch hitters, and prevent 30 runs a season by limiting the number of times through the order by your starter (and by enabling your best pitchers to pitch on shorter rest). I think it might be a little optimistic, but sabermetricians even today suggest that this strategy would work.
A pitcher's opponent's OPS goes up 15-20 points every time through the order, so the idea is to limit the number of times through the order for your pitchers.
I think it's worth trying as an experiment by a NL team with mediocre starters and a good deep bullpen. I'm curious to see how it would work "real-world."
April 20th, 2017 at 10:27 PM ^
In large part, that’s because managers are trying to maximize the number of saves for their closer, as opposed to the number of wins for their team. They’re managing to a stat and playing worse baseball as a result.
I hate it at work and I hate it in my beloved baseball - I hate it when people manage to numbers to the point where quality suffers and the number, while looking good, rings hollow. Gossage was correct in the beginning of this article.
Not sure what the correct answer is, but this is an interesting stat proposal.
Unfortunately, the guy who proved you right is both A) dead, and B) forgotten.
https://en.wikipedia.org/wiki/W._Edwards_Deming#Seven_Deadly_Diseases
April 20th, 2017 at 10:57 PM ^
April 21st, 2017 at 12:52 AM ^
Some pitchers prefer to know exactly what inning they're going to be used. Some don't care.
I don't think anyone's mentioned it yet, but the single biggest problem with the save statistic is how it's almost solely used to "value" a relief pitcher come his arbitration years (see the Dellin Betances fiasco earlier this year).
Also, teams will occasionally hold their best reliever out of the closer role during their pre-arbitration years to prevent them from accumulating saves, which in turn supresses thier cost when their arbitration years roll around.