Radatz makes the cut if I lower relief games to 350, along with Turk Lown, Billy Koch, Danys Baez, and Eric Gagne.

-- MWE

Jeff Brantley
Duane Ward
Keith Foulke
Jeff Shaw
Mark Davis
Bryan Harvey
John Hiller

Francisco Cordero qualifies under criteria #1, I think.

Steve Bedrosian?

I should have said that this is through 2008. Cordero didn't qualify on criterion #1 prior to 2008 (he'd pitched only 62% of his career appearances as his team's closing reliever); he will qualify under #2 after I factor in 2009. Bedrosian just misses on criterion #2; he had 398 appearances during the periods in which he was his team's closeout reliever.

Of the guys in #2, Harvey had only 322 relief appearances. Brantley, Ward, and Davis all had fewer than 50% of their career appearances in the closeout role (Ward was only a closer for one season, then he got hurt). Foulke, Shaw, and Hiller all hard fewer than 2/3 of their relief appearances and fewer than 400 total appearances as their team's closeout reliever (although Foulke is close on criterion #2). There's enough uncertainty in the way that I try to determine the team's closeout reliever (as we know the definition has shifted over time), and their teams did enough sharing of the role, that I might add Bedrosian and Foulke to the list, but the other guys miss by more - Hiller, for example, had 297 appearances out of 502 in the closeout role (59.1%) - and they played for teams that did less sharing of the role.

-- MWE

I put criterion #2 in to catch some of the long-term relievers who had significant closing stretches but who pitched a lot in other roles well. Hoyt Wilhelm, Don McMahon, and Lindy McDaniel, for example, don't qualify under criterion #1 because they all had lengthy stretches late in their careers where they weren't being used in important late-game situations. (I'm also missing significant chunks of McMahon's career because Milwaukee is one of the teams most affected by missing Retrosheet games.)

-- MWE

I put criterion #2 in to catch some of the long-term relievers who had significant closing stretches but who pitched a lot in other roles well. Hoyt Wilhelm, Don McMahon, and Lindy McDaniel, for example, don't qualify under criterion #1 because they all had lengthy stretches late in their careers where they weren't being used in important late-game situations. (I'm also missing significant chunks of McMahon's career because Milwaukee is one of the teams most affected by missing Retrosheet games.)

This sounds a bit like Cy Young and MVP winner Guillermo (Willie) Hernandez but he doesn't qualify having pitched only 744 games and probably not having pitched half his games in a "primary lead protection role".

Tippy Martinez could not be said to have been the "primary closer" in '78-'80, but led his teams in saves '75-'77 and '81-'84.

337 of his 544 relief appearances (62%) were in seasons where he lead his team in saves. So, he'd be above the line of a graph between your two points -- at (400, .67) and (800, .5).

Thinking about relievers from not just before the notion "closer" came into vogue but also before the big-inning-load ace relievers (Marshall, Gossage, Lyle, et al.) had their heydey, I decided to track down Joe Hoerner. It turns out that he had only 498 appearances total, and he transitioned from ace reliever to proto-LOOGY well before his career ended (and even when he as an ace reliever, he was sometimes co-ace, as with the '67 Cardinals and Ron Willis). So he almost certainly doesn't meet your 2/3 requirement for a 400-game pitcher. In his time, complete games were common enough to cut into the opportunities for lead protection.

I'm surprised Gene Garber didn't make the cut. He is 16th all-time with 609 games finished, got 218 saves. He must have been in too many bullpens were it was split time for closing duties.

Mike--Would it make sense to use differing criteria for, say, pre-1980 relievers? Would an accounting of "last-inning protection" games show that the average per appearance for these pitchers is consistently lower than today because roles were more mixed in that time frame? If you did a year-by-year accounting, what would the averages be like for pitchers with, say, 10 or more saves? If the averages are significantly lower back then (as I suspect will be the case), then you might want to adjust for that in, say, the 1954-1979 time frame.

Mike--Would it make sense to use differing criteria for, say, pre-1980 relievers?

Not really. One of the things that I've learned while looking at bullpen usage is that the role of *closer* was well-established by 1960, and that there have really been only two significant evolutions to that role. The first, which began in the mid-70s, was the gradual decision to stop using the *closer* in other roles (specifically in close games when the team didn't have the lead) and the second, which began in the mid-80s, was the gradual decision to save the closer for situations that occurred later in the game.

I should emphasize (again) that I didn't make these decisions by looking at save totals, but by how and when the pitchers were actually used. Tippy Martinez doesn't qualify as a closer in 1975 because he got only occasional late-inning chances; he just happened to convert most of them, where the nominal closers (Lyle and Tidrow) weren't. He doesn't qualify as a closer for 1982 because Stoddard was in that role when he was healthy; Martinez closed while Stoddard was out, and while Stoddard was rounding back into form after the first injury he got some opportunities there as well. Save totals aren't always the best way to identify the go-to reliever on a team, at least before about 1990.

-- MWE

Tippy Martinez doesn't qualify as a closer in 1975 because he got only occasional late-inning chances; he just happened to convert most of them, where the nominal closers (Lyle and Tidrow) weren't.

Actually, when I went back and looked at how Martinez was actually used, it was clear that after his first handful of relief appearances he WAS the closer, and I gave him credit for that. But even when I did that, he still was well under the criteria that I used - and that would still be the case even if I gave him credit for his closing period in 1982 as well.

Tippy is now at 544 RA, 241 as closer for his career when I factor in his closing period in 1975 (his last 15 relief appearances). Even if I credited him with all 76 appearances in 1982 (which is dubious - I came up with no more than half of them that would qualify when I looked at his usage pattern while Stoddard was out and while Stoddard was working his way back to health) that would take him to 317, which is still below 60% of his career total of appearances.

-- MWE