Sunday, April 28, 2013

A Save Was a Save Was a -- But No, It Wasn't

Don't ask me why, but this morning I was looking at the New York Times obituary of Brooklyn Dodgers pitcher Jim Hughes, who died in 2001 at the age of 78. The headline identified Hughes as "Relief Pitcher Who Set Dodger Mark for Saves." Despite the title of this post, the headline contained not one but two misconceptions which continually piss me off, but I will discuss only one of them here.

Jim Hughes did not set a franchise record with 24 saves in 1954, though the obituary writer, the estimable Richard Goldstein, claimed that his total of 24 led the major leagues that year. Saves did not exist in 1954. Even Chicago sportswriter Jerome Holtzman, who created the "save," would have been stumped if you had said to him, following the 1954 season, "Hey Jerry, how about that Jim Hughes! He led the majors with 24 saves this season." It would have been as incomprehensible to Holtzman as informing him that Willie Mays had led the National League in BFW, WAR, and oRAR, not to mention Total Zone Runs, but was only third in Base-Out Runs Added. Those numbers were not calculated in 1954 either. The only difference is that Saves became an official stat in 1969, while the others are more recent sabermetric concoctions.

How can you lead the league in something that doesn't exist? This isn't a simple counting statistic like RBI, which was not an official statistic until 1920. It is reasonable enough to go back through pre-1920 games and calculate the number of runs each batter drove in, but when I write about pre-1920 players, I try to be careful to say "in 1918, Sherry Magee topped the National League by driving in 78 runs." He drove them in whether anybody was counting them or not. Later on, people came up with a stat they named "runs batted in" and figured out that Magee was the 1918 leader. But it isn't the same with saves. Nobody in 1954 was counting saves, because they weren't countable. You had to come up with criteria to measure what you wanted to measure, and devise a formula or rules to apply to those criteria, before you could have something called a save. It took more than two decades after Jim Hughes set a franchise record by pitching in 60 games in 1954 for the powers-that-be to determine what a "save" ought to be, at least to the satisfaction of statisticians measuring the past 39 seasons.

By coincidence, the year the save became official, 1969, also marked the publication of the first Macmillan encyclopedia, the first such tome to include comprehensive statistical data. One thing that the Macmillan editors did was go back and retroactively assign saves for all pre-1969 seasons, using the original criteria proposed by Holtzman and adopted by MLB. Those rules were more liberal than today's. They remained in force for four seasons, and in 1973-1974 a much stricter set of criteria were adopted. In 1975, a middle ground was found, and the 1975 criteria have remained unchanged since then. In other words, we are currently in the 45th season of the save as an official stat, and EVERYTHING you read about pre-1969 "saves" is based on criteria that existed for only four of those seasons.

That is the problem, and to me it's a huge one. The only accurate thing Richard Goldstein could have written about Jim Hughes' 1954 "save" total is that "according to rules in force in 1969 when saves were calculated retroactively, Jim Hughes had 24 in 1954, the highest total in the major leagues." That doesn't sound very convincing, does it? It doesn't have the force of declaring definitively that he led the majors. A compromise way of saying it would be that "Hughes saved more games in 1954 than any pitcher in the majors," the equivalent of my statement that Sherry Magee drove in more runs than anybody else in the National League in 1918.

The most accurate statement you can make--and the one I always make when I can say so with certainty--is "according to today's criteria, in 1954 Jim Hughes blah blah blah." As a matter of fact, according to today's criteria, Hughes would have had only 18 saves in 1954. Not only would that not have been the top total in the majors, it wouldn't even have led the National League. Further, Goldstein could not have written in the obituary that the "record" held up for 35 years, tied by Jim Brewer in 1970 and was finally broken in 1989 when Jay Howell had 28.

Most writers utilize the save stats available at and, the latter website having imported the Retrosheet data. A year ago, I asked Dave Smith, who has done more for baseball historians than anyone I know of by posting box scores and play-by-play data for more than six decades of games, to post a caveat on the Retrosheet website advising visitors that the save data presented therein is based on the 1969 rules and not today's. I just looked around the site and couldn't find any such disclaimer. However,, using the same data, does present a history of saves and save rules, ending with this statement:

"It was possible, under both earlier versions of the save rule, to see boxscores in which pitchers were credited with saves in situations where they would not earn them under the current rule. See for example the game of April 25, 1970, where Claude Raymond entered the game with a four-run lead in the ninth but was awarded a save anyway. For games played before 1969, saves have been figured retroactively using the current definition, and there is no such discrepancy."

That statement is incorrect. The pre-1969 games have not been recalculated using the current criteria. Baseball-ref presents the same data as Retrosheet, based on the 1969 rules. The key difference in the rules is that in 1969, all a reliever had to do was enter with a lead and record the last out of a game without relinquishing the lead. Today we wink at a pitcher who slides in under the current rule's most lenient definition, namely pitching three innings with any kind of a lead; enter in the seventh inning with your team ahead, 15-0, and you can get a save. Well, in 1969 you could enter in the ninth inning with two outs and a 15-0 lead, get that last out, and be handed a save.

From a semantic standpoint, this made no sense at all, which was the reason it was eliminated in the 1973 rule change. To "save" something or someone, there must be peril involved. The award for relievers used to be called the "Fireman of the Year," implying that the game was truly in danger and might have been lost if the reliever hadn't "saved" the day. The weakest 1969 criterion lives on in the three-innings-no-matter-how-huge-the-lead loophole, and even the rule in force since 1975 applies a good deal of lenience. The key provision is that when a reliever enters, the tying run must be on base, at bat, or in the on-deck circle. So you can enter with two outs in the ninth inning with a two-run lead and nobody on base, a three-run lead and one runner, or a four-run lead with two runners. I can understand the reasoning. The most important batter a reliever faces is the first one. If that first runner gets on base in the above scenarios, the reliever is suddenly facing the tying run at the plate. That isn't much leeway for getting your feet (or the ball) wet out there on the mound with the game on the line. Incidentally, in the rule version in force in 1973-1974, a reliever had to enter with the tying run on base or at the plate; in other words, Mariano Rivera couldn't get a save by coming in to pitch the ninth inning with a two-run lead.

Let me use the example of Jim Hughes' 1954 season to illustrate what I mean about how what you read about pre-1969 "saves" is so misleading. Of the 24 saves assigned to him by Macmillan, Retrosheet, and Baseball-ref, six would not have been saves in 1975 or 2013. Here they are:

  1. June 5, Wrigley Field: Hughes entered in the eighth inning with the Dodgers ahead 8-3, two outs, and two runners on base. The potential tying run was somewhere in the dugout. He struck out Ernie Banks to end the inning and retired the Cubs in the ninth for the 8-3 win.
  2. June 19, Ebbets Field: Hughes entered in the ninth inning with the Dodgers ahead of the Cubs, 6-0, one out, and a runner on second. Ooooh, scary! The first batter he faced, Ralph Kiner, belted a two run homer, and Hughes still wouldn't have been in a "save situation" in 2013. He got the last two outs for an easy 6-2 win.
  3. June 27, Ebbets Field: Hughes entered in the eighth inning with the Dodgers ahead 8-3, nobody out, and two Cardinals on base. Once again he surrendered a home run to the first batter he faced, Ray Jablonski. He still coasted to an 8-6 victory.
  4. July 5, Forbes Field: Hughes entered in the ninth inning with no outs and the bases loaded. He gave up a two-run double to the first batter he faced, but don't worry. The Dodgers still led, 8-4. Hughes allowed two more runs to score before securing the final out of an 8-6 win.
  5. August 25, Crosley Field: I warned you this might happen. Hughes entered in the eighth inning with a 13-2 lead. Without the pressure of the potential tying run tying his shoelaces in the dugout, he retired six straight Redlegs to close out the romp.
  6. September 15, Ebbets Field: Hughes entered in the eighth inning with the Dodgers leading, 8-2, no outs, and two runners on base. The Dodgers won, 10-4. 
Those efforts were good enough to pad Hughes' retroactive total in 1969, but in reality he only "saved" 18 games.  That did not lead the National League. Cincinnati's Frank Smith, credited with 20 saves by the usual suspects, had only one which would be disqualified today, leaving him with 19, one more than Hughes. Both would be topped by Johnny Sain's 20 for the Yankees, marked down from 22. Hughes' 18 would still have been the franchise's top total until Jim Brewer's 24 in 1970. In case you're wondering, when I write about pitchers' save totals from 1969-1974, I do not recalculate them according to today's rules. The stat was official and those were the rules at the time, so I don't monkey with them. But it's open season on the pre-1969 saves, and I think it's far more legitimate to apply a 39-year-old (and counting) set of criteria than a set that was justifiably discarded after just four seasons. 

Dave Smith was right, I believe, to justify Retrosheet's data to me when I pressed him on it a year ago.  People went to a lot of trouble to accumulate that retroactive data for the 1969 encyclopedia, and it would take nearly as much effort to recalculate them now. The computer formula would be tricky to write and to apply uniformly to game data going back 100 years or more, but the computer could do it. I could do it myself, one box score at a time, for all the seasons for which Retrosheet provides play-by-play data--currently back into the 1940s, and maybe when I'm retired I will. I've gone through over 100,000 box scores on Retrosheet before for a little research project, and I could do it again. In the meantime, I wish Dave Smith would post a disclaimer at and that Sean Forman would correct the disclaimer at 

I also wish you would spread the word that all published statements regarding pre-1969 "saves" should be taken with an iceberg of salt. Do not believe it when someone writes that Ed Walsh led the American League in saves five times from 1907-1912. He didn't. How can you lead the league in a formulaic stat that did not exist during your long lifetime? Of course, everybody knows that Walsh led the league in "quality starts"--but "saves"? Give me a break.


Cliff Blau said...

A. If there was no such thing as a save in 1954, why did I read a 1952 NY Times article that said Ralph Branca had saved seven games in relief in 1951?

B. Runs Batted In depends on how it is defined as much as the save rule does. If you ground into a double play and a run scores, do you get an RBI? If you did it in 1920 you did, but if you did it in 2012 you didn't.

C. The second save rule was in effect only in 1974: see the February 2, 1974 Sporting News, p. 37, for example, for proof. Actually, it's obvious if you just look at the number of saves in 1972, 1973, and 1974.

Lisa said...

Either way...over 10 years since his death and 50+ years after his last major league appearance, you're still talking about him, so "saves" or not, he was still a great player.