Game Changers: How to save the save rule
Win Probability Added factors context of reliever's appearance into measurement
Before we get going here, I'd like to beg forgiveness from Mariano Rivera, Trevor Hoffman and Lee Smith, among other retired greats. Also Mark Melancon, Craig Kimbrel and Trevor Rosenthal, and most of all the late Jerome Holtzman, my former Chicago Tribune colleague, who pioneered the statistic that has been used to identify the best relief pitchers.
None of them would like what's coming next.
As it's written, the save rule no longer works, and it's time to rewrite it. My solution? Use Win Probabability Added (WPA). More on that in a moment.
Point is, there are better ways to evaluate a reliever's performance than by giving the glory to the guy who gets the last three outs, especially in games where the heavy lifting happened in the eighth inning. Or the seventh inning. Even the sixth inning.
A.J. Hinch, the brilliant manager of the Astros, confirmed my feelings about eliminating the save in a conversation at the Winter Meetings.
"The save stat, I know, impacts how people are compensated and whatnot,'' Hinch said. "But in a perfect world, I wouldn't have it.''
A seventh- or eighth-inning hold -- possibly earned by getting big outs against the middle of an opponent's order -- doesn't move the statistical needle like a 1-2-3 ninth inning with a three-run lead against the bottom of a team's order. So managers -- who know that reliever status and salary have long been dictated by the save -- are understandably reluctant to use their closer in situations that don't fit the rule.
How saves lost meaning
Holtzman introduced the concept of a save in his Sporting News columns in 1960. He was concerned that wins weren't a true measure of a reliever's success -- Elroy Face had gone 18-1 for the '59 Pirates in part by frequently allowing the tying run to score -- and that reliable relievers like the Cubs' Don Elston weren't properly rewarded (or paid) because their work wasn't truly quantified when measured merely by games worked, innings pitched, wins, losses and ERA.
Holtzman devised save rules and tracked them himself for years in the Sporting News before Major League Baseball implemented the rule in 1969. Interestingly, Holtzman's original rules did not require a reliever to finish the game to earn a save, only saying that only one could be awarded per game. But the rules adopted by MLB did require the reliever to get the final out to earn a save.
Consider how the significance of a save has been lessened through the years. The single-season record has been broken or tied five times since 1983. John Hiller, who became the single-season leader with 38 saves in 1973, worked more than an inning 39 times in 65 appearances that season, piling up 125 1/3 innings. Bobby Thigpen worked more than one inning 16 times en route to his 57 saves in 1990, with 47 of his saves coming in outings of one inning or less.
Francisco Rodriguez broke Thigpen's record with the Angels in 2008, saving 62 games. He never worked more than one inning, and manager Mike Scioscia never put him into the game earlier than the ninth. That's the template for the modern closer. In 2015, saves leader Melancon worked more than one inning in only two of his 78 appearances.
Managers are managing to the stat, and that's the tail wagging the dog. Let's overhaul the rule before it becomes completely irrelevant.
We can easily do that because of Win Probability Added, which has become accepted as the ultimate context measurement for the value of a hit or an out during the course of a game.
Let's change the save rule to say that in the games that generate a save under the current rules (margin of three runs or fewer), the save will now go to the reliever not getting the win who has the highest WPA, not the guy who gets the final out.
In the 2015 postseason, there were saves generated in 19 of the 36 games played. Using WPA as the deciding measurement, the save would have changed in 10 of those 19 games.
Among the examples of setup men who would have been rewarded with saves:
• The Rangers' Jake Diekman, who worked a perfect seventh and eighth inning with a 5-3 lead in Game 1 of the American League Division Series against the Blue Jays (instead of Sam Dyson, who pitched a scoreless ninth).
• Diekman again in Game 2 of the ALDS, when he pitched a perfect ninth and 10th to preserve a 4-4 tie (instead of Ross Ohlendorf, who worked a scoreless 14th inning with the Rangers leading, 6-4).
• The Royals' Kelvin Herrera, who worked a four-batter eighth inning with a 5-3 lead over the Blue Jays in Game 2 of the AL Championship Series (instead of Wade Davis, who protected a 6-3 lead in the ninth).
Under the current rules, saves can be simple -- working one inning with a three-run lead or getting one out if the on-deck batter represents the tying run -- or incredibly difficult. The Mets' Jeurys Familia wandered into a real nightmare in the deciding fifth game of the World Series.
Manager Terry Collins stuck with Matt Harvey until the tying run was on second base with no outs in the ninth inning. Familia got three straight groundouts but still wound up with his record third blown save of the Series.
Using WPA to award saves, Familia still would have led postseason relievers with four saves, followed by Davis with three. But Diekman and the Astros' Tony Sipp would have had two each. Seems fair to me.
Another solution: Punt the rule
But here are the problems with using WPA to award a save: Nobody would know what was the save situation until the game was over, and the guy who allowed a ninth-inning lead to get away would take the blown save every time, while he would earn a save only some of the times that he protected a lead.
So another option is to get rid of the save completely. While the idea of killing a stat sounds radical, it has happened before.
"The game-winning RBI,'' said John Thorn, MLB's official historian. "It was created, it was tracked and then we found it wasn't especially meaningful, so it was eliminated.''
How would we measure relief pitching without a save? That's easy. We'd use the cumulative Win Probablility Added stats, which don't differentiate between closers and setup men.
Consider how it might change our view on the 2015 season.
The top 10 in saves: Melancon, Rosenthal, Familia, Brad Boxberger, Huston Street, Kimbrel, Rodriguez, Santiago Casilla, Andrew Miller, Kenley Jansen and Zach Britton. (Miller, Jansen and Britton each had 36, which tied for eighth.)
The top 10 in WPA: Melancon, Dellin Betances, Miller, Davis, Tony Watson, Shawn Tolleson, Brad Ziegler, Familia, Hector Rondon and Kimbrel.
Seven of the WPA leaders are closers who are in the top 10 (and the total grows to eight if you include Davis, who replaced Greg Holland as the Royals' closer in August). But sprinkled among them are Betances and Watson, who combined to enter only 15 games in the ninth inning or later.
Only four of the top 10 in saves were also in the top 10 in WPA, and is anyone going to argue that those guys (Melancon, Familia, Kimbrel and Miller) aren't among the game's best relievers? The most prominent omissions were Rosenthal and Boxberger, with those two finishing 14th and 129th, respectively, in WPA.
Old-school thinkers might scoff at WPA. That's OK. It is the product of a complicated formula. But think of it as MLB's version of Quarterback Rating. When the National Football League introduced that stat in 1973, few accepted it as the best measurement to rate passers. But that has changed as generations of fans have known the game only with QR in the landscape.
The save was a useful tool when Holtzman introduced it and the Elias Sports Bureau refined it, but now it has become an unnecessary set of handcuffs for managers.
It's time to do them a solid. Let's remove the cuffs and watch what happens.