Just doing some of my own calculations on Chapman, and the leverage he's been in I find his impact on W/L much greater than WAR.
Compared to a .5 ERA, based on stochastic simulation, he has .4 more wins
Compared to a 1.5 ERA, the team has 1.2 more wins
Compared to a 2.5 ERA, the team has 1.9 more wins
Compared to a 3.5 ERA, the team has 2.6 more wins
And compared to replacement level the team has 3.5 more wins, where as
WAR values him at 1.8, and points to statistical problems when trying to measure the extremes with averages.
On a similair note, during Mariano Rivera's career as a closer(did not include this year - not using incompletes, or his first year - not yet a finisher)
The Yankees in that time frame have averaged 3 wins / year above Pythag, if you credit that due to leverage as opposed to luck (which I believe) it turns Rivera into a 6.5 Win/Year player. Also of note, the Yankees worst year compared to Pythag is 2007, the only year as a closer Rivera had an ERA over 3. The Yankees best year compared to Pythag, also happened to be Rivera's highest save output - 2004 (+12 compared to Pythag)
The 5 years Eckersly averaged under a 3 ERA as a closer the A's averaged +4.4 over Pythag.
Between 1996 and 2007, the Padres only had one year below Pythag- 2003 the year Trevor Hoffman was injured, they were also below in 1995 and 2008, the two years with a 3+ ERA as a closer All in all only 1 in 32 years of these closers did the closer have a era lower than 3 and the team finished below Pythag, so is above below Pythag more luck, or more skill of a closer?