Am I right in thinking that WARP is more-or-less based off ERA? This being true, the offhand remark that "WARP treats a scoreless ninth the same as a scoreless first, so it gives Rivera no extra credit for his high-leverage outings" would seem to dismiss a critical factor.

Would it make sense to multiply pitchers' WARP by the leverage value for the situation they inherit?

It seems like the answer is linked to your comment:

"And WARP treats a scoreless ninth the same as a scoreless first, so it gives Rivera no extra credit for his high-leverage outings. "

Well, that would suggest that WARP alone is a flawed way to make the "value" evaluation. Rivera's leverage adjusted WARP will compare favorably to starters who pitch, on average, in dramatically lower leverage situations.

The fact that WARP values Pettitte's career at 20 wins more than Rivera's should raise questions among its acolytes that maybe WARP isn't an ideal metric for relievers.

Come on, people. We know that the idea of double and triple counting higher leverage innings just because of the way that modern bullpens are constructed is nonsense. Rivera wouldn't have been a significantly worse pitcher if he'd always pitched the 6th, 7th or 8th inning. Relievers are less valuable than starters because they pitch so many fewer innings - it's really that simple.

How has John Smoltz's name not entered this discussion? He was to a great extent contemporary with Rivera and was both an elite starter and an elite reliever.

Rivera was probably a slightly better closer than Smoltz, so if you don't mind extrapolating from one data point, you'd guess that Rivera would translate to a slightly better starter.

Of course, this whole analysis (mine as well as the one above) is predicated on the assumption that "starter" and "reliever" is the same job. To me, that's like saying "outfielder" and "shortstop" are the same job because they're both come to the plate and hit the ball. A 2.50 ERA from a starter is different from a 2.50 ERA from a reliever in the same way that an .850 OPS form a 1B is different from a .850 OPS from a SS. For position players, we make adjustments to their WARP based on their position. It seems that we haven't found an intelligent way to make a similar adjustment for pitchers, though.

No reliever belongs in the hall of fame. They really messed it up letting Sutter in.

I saw Dick Raditz pitch when he was with the Red Sox from 1962-1966. His "greatest year" was 1963, not 1964. Raditz began to be more hittable in 1964 (and I don't just mean John Callison's All Star Game winning homer off of the "Monster")
We are talking about a much different era in baseball history,
although by 64 most teams had at least one pitcher to close out games. The great thing about Radatz was his control and his pitching motion which made it difficult for hitters to pick up his fastball

It may be a little simplistic but Rivera has the second highest career WPA of any pitcher, behind only Clemens. The rest of the top twelve goes: Johnson, Maddux, Martinez, Mussina, Smoltz, Halladay, Schilling, Hoffman, Gossage, Eckersley. Seems on its face to be a pretty fair ranking.

Is there any evidence that throwing fastballs are more strenuous in the long-term than throwing cutters as you purely speculate?

The most stunning thing about this article is that Carlos Marmol has two of the best seasons by a reliever over the last 10 years.

I'd like to find out what the Yankees record is going into the 9th in a save situation over the Rivera era compared to other teams. Then we'd have a better idea just how monumental the task of replacing is.