Wednesday, September 19, 2012

Luckiest Team in Baseball

This one will be way out there people who normally look at my blog.  I'm a baseball junkie.  A stat geek.  I've been playing a baseball simulation game called Strat-O-Matic baseball since I was a teenager.  I once replayed an entire MLB season using "Strat" on my own.  It was the 1978 season, and there were 2102 regular season games, if I recall correctly.  I played them all.  Plus the playoffs.  At the end, the Yankees still beat the Dodgers in the series, so my 1000+ hours of effort failed to overturn the results of that series.  I think I only re-played the season because I was so disappointed when the Yankees won.  I wanted to reverse it, if just on a personal level.  No luck.  Or given how the game of Strat is played, maybe I should say "no dice."

Today, 35 years later, I still play in Strat leagues.  I'm currently in the playoffs in one league, one win away from a world series appearance.

I was once a member of the Society for American Baseball Research, a collection of fellow baseball geeks.  In SABR, there seemed, at the time, to be two kinds of members - the stat geeks and the history geeks.  And, we needed each other.

What I want to address today is the 2012 Baltimore Orioles.  They are tied for first place in the AL East with the Yankees.  And, they have absolutely no business being there.  In reality, they should be just ahead of the Red Sox, well out of playoff contention.  Yet, they are winning one run games at a ridiculous rate.  They are 27-8 in one run games.

Over history, most teams win one run games at a rate somewhere between their winning percentage in games not decided by one run and .500.  That is, both good teams and bad teams will somewhat regress toward the mean in one run games.  (I hope I don't get any statistical terms wrong here, because one of my readers is a math professor).  Think about it.  If you are a bad team and you lose a lot of games, when you are lucky enough to be in close game, you will probably win more of those than you would win if the game hadn't been close.  Conversely, good teams will lose more often in tight games, because they've allowed an inferior opponent time to stay close and sometimes get lucky.

The same concepts apply in extra inning games, which are quite often 1 run games.  The Orioles are 14-2 in extra inning games.

Ignoring the one run games, the Orioles are essentially a .500 team.  So, you'd expect them to be close to .500 in one run games.  In that case, their record would be 75-73 or 74-74.  Yet, they are 84-64.

Another formula used to project baseball winning percentages, one that assumes a fairly normal run distribution over time, is called the Pythagorean Theorem of Baseball.  I'm not going to go into details here, but that formula focuses on total runs allowed and total runs scored.  Amazingly, the 84-64 Orioles have been outscored this season by a slight margin of 660-652.  According to those numbers and the PToB, they should have won 49.3% of their games, for a record of 73-75.  In other words, they've won 11 more games than expected based on this theorem.  Looking above, they've won 9 or 10 games more than they should have won.

Three other teams are at +5 this year, and one team is -6, but no one is close to +11.  None of the other lucky teams appears playoff bound.  One thing that's funny is that two of the unluckier teams are the Reds and Giants, who despite being -5 are both in first place by decent margins.  In reality, they should have wrapped up their divisions by now.

Yes, there are confounders in this formula.  The first one that occurs to me is an absolutely dominant closer. Great closers are used primarily in close games, and if they dominate, they can push their team's record in close games to higher than otherwise expected levels.

The Oriole's closer leads MLB in saves.  And he's having a very good year.  But, statistically, hes' not as good a pitcher as his set-up man.  (Maybe the set-up man being so good is a boon to the O's in tight games).

Other closers in the league like Fernando Rodney, Craig Kimbrel, Jason Motte, Aroldis Chapman, and Kenley Jansen have clearly been better.  Another half dozen are of similar quality.

So, I don't see a truly dominant closer as the reason for Baltimore winning so many games.  I think it might explain Tampa Bay being 5 games better than projected, but I see no such shift for the Orioles.

I haven't examined historical records for how these O's would rank in the history of good luck, because I can't find them on the web.

I also haven't looked at the results that I would have found using a slightly revised version of the theorem, that uses a different exponent that makes the formula look less "Pythagorean".

However, there is no doubt in my mind that the 2012 Orioles are an absurdly lucky team.  If they take this season and assume that they are actually a good team, and they make few changes in the off-season, fans should expect a significant drop-off next year.  Historically, teams that win lots of one run games tend to get worse the next season, when their luck, essentially, doesn't repeat.  The free agent era has changed this a bit, but holding tight after a lucky season is not the way to repeat.

I would bet that they are one of the 20 luckiest teams in MLB history to this point, but I can't verify that.

But, like most math geeks, I enjoy looking at an interesting statistical anomaly, and the 2012 Orioles are the definition of an anomaly.

No comments: