Guest Post: Girardi and Pythagorean Winning Percentage Or What Happened to The Yankees in 1986?

(** Editor's Note**- This guest post comes courtesy of commenter and friend of the blog Professor Longnose.  It is about historical Pythagorean win expectancy and what it might say about Joe Girardi's managerial skills.) As manager of the Yankees, Joe Girardi has his boosters and his detractors, which is no surprise.  Few Yankee fans have been happy with the results of the last 2 years, but there are arguments about how much of the blame is Girardi’s. One of the arguments I hear quite a bit is that the teams Girardi has managed weren’t very good, and were hit with injuries on top of that, so that even being in the hunt for playoff spots shows how good a job Girardi did. This is more or less the “company line” among Yankee broadcasters.

One stat used to promote this view is that the Yankees in 2013 and 2014 have outperformed their Pythagorean winning percentage. In 2014 they were 7 games better than the Pythagorean formula suggested they should have been, and in 2013 they were 6 games better. That’s pretty impressive; the Yankees were tied for best in MLB with the Cardinals in 2014 in beating Pythagorean expectations, and tied for second best in MLB in 2013. As a twitter buddy of mine said, “That’s a trend.”

Actually, in 2014 they were 6.5 games better than Pythagorean expectations according to FanGraphs, and the Cardinals were over 7, but it rounded to a tie in the baseball-reference standings, which heads this particular column “Luck.” You’ll get an argument about the accuracy of that heading from sabermetricians (here’s one: http://www.fangraphs.com/library/principles/expected-wins-and-losses/), but the stat does represent things that we can’t quantify, at least at the moment.

There are theories. One that makes a lot of sense says that teams with good closers and back-end bullpens will outperform their Pythagoreans because they will win a high percentage of 1-run games. That helped the Yankees in 2014. They were 28-24 in 1-run games, good for a .538 winning percentage, better than their .518 overall percentage, and much better than their .475 Pythagorean percentage. But they played .509 ball in non-1-run games, still a lot better than their Pythagorean. In 2013 it was dramatic. They were 30-16 in 1-run games, a .652 winning percentage, compared with .525 overall and a .488 Pythagorean. In non-1-run games, they played .474 ball. The 1-run games were the difference. That’s pretty good: change the column heading from “Luck” to “Manager’s Handling of the Bullpen” and Girardi looks like a genius.

Another theory is that inconsistent offenses and pitching will have significant effects. A team with inconsistent offense may overperform their Pythagorean, because when they’re hitting they’ll run up the team’s runs scored without helping the winning percentage. On the other hand, teams with inconsistent pitching and defense may do the opposite: when the pitching stinks, they’ll give up a lot of runs that will make the Pythagorean look bad but won’t contribute any more losses than losing a close game.

That doesn’t seem to have had much effect on the Yankees in 2013. They lost 4 games by 8 or more runs (that’s a rather arbitrary standard, but I didn’t have anything else to go on), by scores of 9-1, 12-2, 11-1, and 11-3, being outscored in those 4 games by 36 runs. But they won 3 games by 8 or more runs, 14-1, 11-3, and 9-1, outscoring their opponents by 29 runs. So in blowout games, they were 3-4 and outscored by 7 runs. That doesn’t seem like much. (I remember that 14-1 blowout of Cleveland. It came early in the season and suggested that the Yankees had mined gold in a few scrapheap pickups. It didn’t last.)

In 2014, however, it did have an effect, a big one. They won only 2 games by 8 or more runs: 10-2 and 14-5, but they lost 6, and were outscored in those games 72-12. They were outscored by 43 runs in those 8 games, more than their total run differential for the year, which was -31. In fact, they were outscored by 27 runs in just the 2 worst losses of the year: 16-1 and 13-1. That’s getting pounded. Without their 3 worst losses, their run differential on the year would have been +5 and they would have beaten their expected Pythagorean wins by 2 games instead of 7.

That isn’t entirely fair, certainly not in evaluating the club, which did give up those runs; you can’t just ignore them. It only suggests a couple of factors that may have affected their Pythagorean overperformance in 2013 and 2014.

From all of this, I don’t conclude much of anything. This doesn’t come close to rising to the level of a study; someone should investigate this seriously rather than just picking a few numbers from the last 2 years of his favorite team.

But even though I still didn’t know how much of the 2013 and 2014 overperformance to attribute to Girardi, I wondered how the Yankees have done in Pythagorean expectations in the previous Girardi years. If you assume that this is a repeatable trait of Girardi’s managing, he should be able to do it most of the time. But the Yankees’ “Luck/Managerial Brilliance” totals for 2008-2012 were mixed:

2008: +2 2009: +8 2010: -2 2011: -4 2012: 0

Add in 2013 and 2014’s +7 and +6 and it looks pretty good: a total of +15 over, or an average of +2.4.

At least, it looked good compared with nothing. I wasn’t about to compare numbers for all of MLB for any of these years, but I decided to look back at the Joe Torre era. And compared with Torre, Girardi doesn’t look so great.

Torre managed the Yankees for 12 years and overperformed his Pythagorean in 10 of those years:

1996: +4 1997: -4 1998: +6 1999: +2 2000: +2 2001: +6 2002: +4 2003: +5 2004: +12 2005: +5 2006: +2 2007: -3

That’s pretty amazing. 8 of those years the Yankees were in the top 4 in MLB in exceeding Pythagorean wins. Could they have been really lucky? Was Joe Torre even more of a genius than Girardi?

But what about before Torre? Maybe this was a Yankee thing, or a Yankee Stadium thing. I wound up looking back to 1969, which was the year divisions were introduced and the mound lowered; going back before then didn’t seem relevant.

And here’s what I found: in 1986, aliens came down from outer space and magically granted the Yankees the ability to exceed their Pythagorean expectations nearly every year. The totals for 1986 to 1995, the last year before Torre took over, look like this:

1986: +3 1987: +5 1988: +2 1989: +3 1990: +2 1991:  0 1992: -4 1993: +5 1994: +2 1995: +1

In the 24 years from 1986 to 2009, the Yankees had negative “luck” 3 times.

How about before then? From 1976, when YSII opened, through 1985, they bounced around like you might expect, usually staying pretty close to 0. The numbers:

1976: 0 1977: +1 1978: +1 1979: +3 1980: +6 1981: -4 1982: -1 1983: +3 1984: -2 1985: -1

Overall, that’s a positive record but nothing that seems out of the ordinary.

What about Shea Stadium, where the Yankees played for two seasons while YS was being remodeled?

1974: +3 1975: -8

That doesn’t tell us much. The old Yankee Stadium?

1969: +3 1970: +4 1971: 0 1972: -2 1973: -5

That doesn’t say much of anything either. Of course, Yankee Stadium goes back 50 years before that, and I didn’t look at those numbers.

It’s possible that the Yankees’ run starting in 1986 is just luck, too. Although the odds are greatly against any one team having such a positive streak, the odds are shorter that some team somewhere would have had some streak, and maybe we just happen to be rooting for the team that did.

By now, my goalposts have been moved. I started out hoping that Pythagorean win expectancy would help explain Girardi’s managing. Instead, I found that I need something to help explain Pythagorean win expectancy. I hope someone finds it, or maybe it’s already been found but I’m just not aware of it.

In the meantime, I’m going with “Pythagorean wins don’t suggest that Girardi was a great manager,” and “Aliens visited Yankee Stadium in 1986.” If anyone knows of any studies of either of these hypotheses, please let me know.