Mariano Rivera is now baseball career saves leader, but is it easy to compare him to his non-reliever peers?
Among yesterday's hysterics following Mariano Rivera closing his 602 game, an article was penned by ESPN New York's Wallace Matthews stating that the Yankee closer was the greatest Yankee of this generation. He muses,
"I realize that many will lean toward Derek Jeter...as the greatest Yankee of our era...And it is pretty much impossible to quantify the relative worth of a closer versus a shortstop..."
Impossible? Hardly. Difficult? Absolutely.
Though, it certainly isn't as simple as looking at both Jeter and Rivera's career Wins Above Replacement Player. Evaluating relief pitchers has given WAR(P) a significantly harder than it has with position players. For those who are unaware, in addition to summarizing a player's total contributions to their team in one statistic, Wins Above Replacement (WAR or WARP) removes the context said contribution and focus solely on what an individual player controls and accomplishes.
But, what is a relief pitcher without the surrounding context of the game? The modern closer, while just a name for a misused relief pitcher, derives much of his value from coming in high leverage situations. If one was to attempt remove the context, as sabrmetrics does with its various WAR constructs for every other position, she wouldn't be getting an accurate representation of a closer's value.
The above discussion begs the question, how much credit should one give to relief pitchers for context they had no part in creating?
Yesterday, Mariano came into a 6-4 game in the bottom of the 9th inning and classically retired three straight hitters to record historic save number 602. Some would argue, that Mariano's performance deserves a leverage adjustment to account for outs he recorded in the 9th inning being generally more important to his team than the outs previously recorded.. Others counter that the Mariano had nothing to do with giving the Yankees a 2 run lead, and his inning should be treated as if he had entered the game in a blowout. This would be a context neutral approach.
It has been suggested, just as Wins Above Replacement shows the difference in difficulty between a shortstop and a designated hitter with a positional adjustment, a similar concept be implemented for relief pitchers. Give relievers credit for when they enter high leverage games and ding them for their low leverage appearances. I graphed a simple version of this for several teams this off-season to evaluate the trust that a team's field manager had in a given reliever. However, the series, entitled "Circle of Trust," just accentuates the issue with this proposed method of using leverage index in a coefficient capacity. The leverage index that pitcher is credited with when he is called into a game is dependent on when his field manager calls for him. While one's leverage index may correlate highly with one's ability, the reliever certainly did nothing to create the situation he was handed..
A balance must be struck between keeping true to context neutral statistics like Wins Above Replacement Player by implementing consistent methodology and tuning said metric to give the reader the best representation of a player's value.
Comparing Rivera to a non-reliever is quite difficult to do whether one applies context an advanced metric or not. For now, let us be content with proclaiming Rivera is the greatest one inning pitcher of all time, and save weighing his contributions against Derek Jeter for another time.
 Typically it is just assumed that 9th inning is the most important, which is why I that closers are "misused". Looking over yesterday's game log, you'll notice that Mariano's plate appearances were not among those with the highest Leverage Index (LI) for the game. In hindsight, brining in Mariano in the 5th inning instead of Corey Wade would have been the best use for the reliever.
 Another issue worth mentioning is that sabermetrics specifically does not care how batters' performance was impacted by his teammates. Specifically, thing about any RBI discussion involving Ryan Howard. He had nothing to do with Rollins, Utley, and Co. getting on-base in front of him. If one was to implement some variation of leverage index as a multiplier on relief pitchers' performance, it would follow that they should make the same modification for hitters and starting pitchers. Give everyone credit for the importance of the situation they been thrust into. But, clearly WAR(P) would no longer be a context neutral statistic.
*Update* I guess this comment wasn't saved... This piece was inspired by Mariano's historic accomplishment, Mr. Matthews, and most notably a very very long twitter conversation with Matt Klaassen, Sky Kalkman, Dan Turkenkopf and Colin Wyers. You should follow all of them.