Albeit, the Metropolitans haven't regressed entirely; Apu's fav-or-ite squadron is still outperforming its RPScore by .024. What's more, the Mets have demonstrated some real improvement of late, improving their rating by .039 since July 1st.
The preseason favorite Washington Nationals, on the other hand, represent the Mets' mirror image. Starting out slow, they righted themselves by the middle of May, conforming rather well to their RPScore expectation. Theirs is a case study in a different type of regression: the positive kind.
While the Nats may have been the better team on paper, the Mets have been luckier. With less than a third of the season remaining, the Mets have a one game lead on the Nationals (they're tied in the loss column). Despite the Nationals' superior RPScore, Elo rating, Pythagorean and BaseRun expectations, my simulator projects a tie in the NL East between the Mets and Nats.
In other words, if the theoretically superior Nationals are to take the division outright, they must play above their theoretically superior talent level.