Don't pay too much attention to the 2014 Mets' run differential

The 2014 Mets' run differential suggests that they were better than their 79-83 record indicates, but a closer look reveals that perhaps it's not the case.

The Mets finished the season with a 79-83 record, but had a positive run differential. According to their Pythagorean record, they "should" have finished 82-80, based on the 629 runs they scored and the 618 runs they allowed. On the surface, this appears to be a reason to feel more optimistic about the Mets' season. They were actually better than their record indicates, and maybe that bad luck will even out next season, right? There's just one problem: more sophisticated record estimators suggest that the Mets "should" have won fewer games than they did.

Jared Diamond of the Wall Street Journal also noticed the Mets' run differential, and he asked Sandy Alderson about it:

The Mets believe in the value of run differential, but they recognize that it can't replace real-life results. In a telephone interview with The Wall Street Journal on Tuesday, general manager Sandy Alderson described run differential as "another measurement" to consider.

"Obviously, it's not wins and losses," he said. "It suggests to us that our record was about right or slightly below. As you look at next season, that's somewhat reassuring because it's not a gap that we have to make up."

"Another measurement to consider" is an apt description of Pythagorean record -- it has the right idea, that a team's win-loss record isn't always indicative of underlying performance -- but Alderson might not want to be so reassured. An alternative measure, Base Runs, suggests that the Mets actually overperformed by a few wins.

Base Runs estimates a team's record based on component statistics, removing the effects of sequencing. In the long run, sequencing is random. A team might spread out ten hits in a game and score only one run, or it might bunch up ten hits in an inning or two and score six runs. This is largely random and Base Runs accounts for it. Research supports that Base Runs is the best estimator around (link via FanGraphs).

To loosely analogize the three concepts, a team's actual win-loss record is like batting average, its Pythagorean record is like on-base percentage, and Base Runs is like wRC+. Both of the latter two are an improvement over actual win-loss record, but Base Runs is the king.

Two publicly-available Base Runs estimators are Baseball Prospectus's Third-Order Wins and FanGraphs' BaseRuns. They estimate that the Mets' record should have been 77-85 and 76-86, respectively. Here is a comparison of the various measures:

So instead of underperforming by three wins, the Mets may have overperformed by two or three wins. They probably didn't "deserve" to win 80+ games in 2014. There are reasons to be optimistic about the 2015 Mets, but the 2014 Mets' run differential is not one of them.