OK, let’s pick up with one of the questions that I was interested in, that started this project.   Have relief pitchers, in fact, become more reliable over time?  Relievers in the 1960s, 1970s. . . .they had brilliant seasons, and then they disappeared.   But how much different is it now?

It is different.  In 1915 there was only one pitcher who qualified for the top-10 list, and he did not repeat, so we enter that as     1      0.   In 1916 there was only one, and he did not repeat, so we enter that, again, as     1      0. 

The total for that decade is   4     0.     For the 1920s, there were 12 pitchers (total) who qualified for the top ten list, only one of whom ever repeated (Firpo Marberry, 1925-1926), so we enter that as    12     1.   This is the total for the decades up through the 1950s:

What that shows, of course, is that the Top-10 relievers lists are filling out over time, and (b) that this is not happening very rapidly.  But it is happening. 

But then when we get to the 1960s, we learn something from doing this (which is such a relief; usually I don’t learn nothin’.)  But this time we do.  This is the "repeating on the list percentage" from 1960 through 2010:

That looks like a pretty definitive pattern, doesn’t it?  Numbers can fool you; sometimes it looks like there is information there, but it’s really just random.  But this looks like a pattern.  The percentage of Top-10 Relief Pitchers repeating on the list increased steadily from 1960 to 2009.  Presumably this happened because the change in usage patterns of relievers reduced the strain on them, allowing them to remain effective. 

In the last decade—and we have only 90 pitchers in this sample, because we have no 2020 data—but in the last decade only 27 of 90 pitchers have repeated on the list, or 30%.  This may have happened because the explosion of short-usage relief pitchers in the last 10 years has made it more difficult to stay on the list.   That’s just speculation, though; I can’t say for sure, at this time, that that’s what is really indicated by that pattern. 

The next thing I wanted to get to is the average performance of a #1 Relief Pitcher, by decade.  This chart summarizes that data:

I once heard from a reader that I had overrated Roy Face’s season in 1959.  Face went 18-1 but, said the reader, "a reliever’s job is to save games, not to win them."  That was in the 1980s.  "Saves" were a relatively new thing, but in this gentleman’s mind, they taken control of the subject.  I tried to explain to him that relief pitchers in 1959 were not used in the same way that they were in the 1980s. 

It’s a natural thing.   We all tend to apply the standards of the present to the past—unless you know better.  I find this syndrome all the time, writing about baseball or crime or politics or whatever.  It’s why I am doing this.  I’m trying to help people understand what things were like in the past, and how we got where we are. 

#1 relievers in the 1950s had an average won-lost record of 13-5.  It wasn’t just Face being 18-1; the records of the top relievers for that decade were 16-7, 11-2, 15-4, 10-6, 10-7, 13-5, 14-5, 12-8, 10-3 and 18-1.   They averaged 17 Saves—13 Wins, 17 Saves.  They were often brought in to tie games, and often brought in when the team was a run or two behind. 

#1 relievers averaged 581 batters faced in the 1970s.  In the last decade the average is less than half of that (273). 

Saves by #1 relievers INCREASED in every decade until the last one, and declined in the last decade—again, an interesting observation although it is too early to know for sure what it means. 

The average ERAs of the #1 relievers have improved in every decade since the 1950s, dropping from 2.53 to 1.29.

The effectiveness levels, reflected in hits/inning and strikeout/walk ratio, have steadily and fairly dramatically improved—but, again, over fewer and fewer innings.  Relief pitchers become more and more effective, because they are asked to do less and less. 

An average #1 reliever’s winning percentage has always been over .600, and the variations in that would not appear to be significant. 

And we come, finally, to the question of who was the greatest reliever of all time.   Not saying my answer is correct; not saying that your answer is correct, either, but I have studied the issue in a systematic manner, and this is what I have.  

First, I gave each reliever "points" for appearing on the list of the Top 10 relievers of the season—10 points for being first, 9 points for being second, 8 points for being third; you know the drill.

Except that, for many of the years from 1915 to 1958, there aren’t 10 pitchers on the list, so you can’t actually do that.   So then I modified the points system so that if there is only a 9-man list for that season, then the #1 man only gets 9 points, the #2 man 8 points, etc.  

Except that that seems unfair to a pitcher who is the top reliever of the year, but is the only one who qualifies for the list.   So then I modified the rule from the previous paragraph, so that the #1 man—if he qualifies for the list, at all—always gets at least 3 points.  

Then you add up the career points for each reliever.   These are the ten top relievers ever, by that system:

Turns out this "Mariano Rivera" guy was pretty good; stop the presses.  In a moment, just for shits and grins, I’ll give you the Top 40, just so you can find Aroldis Chapman and Dave Righetti and Jesse Orosco and Tom Henke and all the other pretenders that some twit on Twitter has tried to tell me was the greatest of all time.  I’m sorry; I guess Righetti and Orosco don’t make the Top 40.

Anyway, if Mariano wasn’t at the top of the list I probably wouldn’t have printed it, but there was something on the Top 10 list that really stunned me.  Joe Nathan is the 4^th greatest reliever of all time?   You’re kidding? 

I had to double-check my work, make sure I hadn’t entered as "6" as "26" or something.  I hadn’t.  I just had not realized that Joe Nathan’s was AS dominant, compared to his time, as he was.  Nathan ranks as the #1 reliever of 2006 and 2009, and ranks in the top five in 2004, 2005, 2007, 2008, and 2013.  Billy Wagner can’t match that, and Rollie Fingers can’t match that, and Trevor Hoffman can’t match that, and Bruce Sutter and Lee Smith aren’t even playing in that league.   If you add Bruce Sutter and Lee Smith together, they rank ahead of Joe Nathan, but just barely.  I was quite stunned to learn this. 

I was so surprised by this that I decided to rank them a different way.  My second ranking was this.  I arranged relief seasons by age, rather than by year, and ranked all relief seasons 1 through 25 by age, from ages 20 to 40.  This is consistent with my often-explained philosophy of research:  look at the issue in different ways, with different sets of rules, so that you can be sure that what you are seeing is a real thing, rather than a chance impression created by the way you happened to do your ranking system. 

So I ranked them 1 through 25, and I gave each pitcher 25 points for finishing first in his age group, 24 points for finishing second, 23 points for finishing third, etc.    The best season ever for a 26-year-old reliever was by Eric Gagne in 2002; the best season ever by a 31-year-old reliever was by Mike Marshall in 1974, etc. 

I put all pitchers who were (a) under age 20, or (b) over age 40 into one age group and took the 25 best in that group.  The best season ever by a reliever over 40 or under 20 was by Hoyt Wilhelm in 1964; the second-best was by Mariano in 2013.  This system works to the advantage of Hoyt Wilhelm, as it enables him to pick up multiple mentions in the same category, which seems fair; Wilhelm was consistently brilliant at an advanced age.   I designed it to be fair to Hoyt Wilhelm. 

Anyway, when I do it this way, Joe Nathan has to compete with Dick Radatz and with Willie Hernandez, 1984, and Lindy McDaniel, 1960, and all of the great relief seasons in history.  And this is the Top 10 relievers ranking which results from that effort:

Joe Nathan is no longer fourth.  He’s now third.  Studied in that way, Mariano holds the top spot and the list shuffles quite a bit, but Joe Nathan moves UP, from fourth place on a list of the greatest relievers ever up to third place.  How about them apples? 

So I concluded, then, that I just never realized how good this guy really was. 

On learning this, I posted on Twitter that Joe Nathan was the most underrated relief pitcher of all time.  Some people agreed with me, and a lot of people tried to tell me it was Ron Perranoski or John Wetteland or Ted Abernathy or somebody, and one delusional gentleman wrote "You misspelled Pat Neshek."  Pat Neshek’s score in one of these systems was zero, and in the other one, zero. 

Anyway, several people pointed out that Nathan’s performance in Post Season play was terrible, which (a) is entirely true, (b) is very relevant to the issue, and (c) I was not aware of.  As I had overlooked Nathan’s excellence, so too I had overlooked his failures in post-season play. 

So maybe he isn’t the third or fourth greatest reliever of all time, because of his post-season failures, or perhaps he isn’t because my ranking system just isn’t that good.   Most people’s ranking systems are what is generally called "intuition", so I’ll put my research ahead of their intuition, but I’m not claiming this is a perfect method.  All I am really saying is, Joe Nathan’s career was one hell of a lot better than I ever realized that it was. 

I promised you top 40 lists, so here goes.  Hall of Famers in Bold Face.

The list on the left, the highest score for one season is 10 points; the list on the right, the highest score is 25 points.  Mariano’s score can thus be represented as 10.8 seasons as baseball’s best reliever (108 divided by 10), or 10.2 seasons as baseball’s best reliever (255 divided by 25.)  Joe Nathan’s score can be represented as 5.7 seasons as baseball’s best reliever (57 divided by 10), or 5.76 seasons (144 divided by 25.)  

Systems like this are intended to inform our opinions or guide them, not to dictate to us.  How you rank the best relievers is up to you; I’m just trying to inform the discussion.   Thanks for reading. 

(Sorry about Roberto Hernandez being on the list twice.  I have tried to fix that, but our posting software has defeated me.  Roberto Hernandez should only be on the list once, and his correct score is 26. I'm not sure what went wrong in the underlying program.)