Methodology
In 2011, the Yankees bullpen had a 3.12 ERA, with the average for all pitching in the AL being a 4.08 ERA. That is 31% better than league average, so the bullpen posted a 131 ERA+. The 2011 Yankees scored 867 runs and allowed 657, which using Pythagorean Winning Percentage translates to an expected 101 wins. They actually won 97 games, which means that they under-performed their expectation by four wins, or a -4 on the graph. I did this for every season from 1990-2011 and put them on a scatter plot.
But what does it all mean, Basil?
A team over-performing their Pythagorean wins indicates winning a lot of close games; they pulled out more wins than the raw run differential would have predicted. Under-performing means the opposite; they lost a disproportionate amount of games, given their run differential.
*One step further: If a team has a proclivity for winning blowouts and losing close games, they will under-perform their Pythagorean expectation, as one 15-0 trouncing carries a lot more weight in run differential than a few 3-2 losses. People say this about the Yankees all the time, but from 1990-2011, they have over-performed their expectation by 46 games.