Rob Neyer wonders:
The Yankees are overwhelming favorites because they're significantly more talented than the Phillies. If they'd played in the National League East this season, they'd have beaten out the Phillies for first place by a dozen games or more.
Which doesn't mean they're going to win the World Series. But there's a reason why so many people who don't live in Philadelphia think the Yankees are going to win. And while I know that the numbers (and many of the same people) believed the Dodgers were going to beat the Phillies, I also know that the numbers (and the people) have done pretty well this month, otherwise.
I would love to see an exciting World Series. I just don't see any reason to think I will.
Take his assertions as true: going by third-order wins or JinAZ's TQI, the Yankees would have won about 13 more games than the Phillies against a neutral schedule.
Given that, is he right about our likelihood of getting an exciting series?
There are two problems with Neyer's line of reasoning, even once we grant all of his assumptions. First, it fails to establish a reference class for what constitutes an exciting series. Second, even given a reasonable reference class, it is unclear that he is correct. Let's take them one at a time.
First, in order to even be able to evaluate whether we will see an exciting World Series, we must establish what would count as one. The maximally exciting series would feature seven extra-inning games with outstanding pitching performances, spectacular defensive plays, and walk-off home runs. Clearly the bar cannot be this high.
Let's say that an exciting series is one that goes six or seven games, with either team winning. The Yankees-Angels series went that long, and I think most would agree it provided a good degree of excitement. Part of that was the up-and-down nature of the individual games, but we can't predict that sort of thing from the ex ante perspective.
Having established an excitement threshold, how likely are we to see it met? Baseball Prospectus, which uses a Monte Carlo simulation and pitcher matchups, has the Yankees winning the series 59.9% of the time. And while we can't see the cross tabs (to borrow a polling term), it is clear that in a significant number of those trials, the series went to six or seven games. It simply cannot be the case that the distribution is two-peaked, with the most likely outcomes being the Yankees in four or five games and the Phillies in four or five games. (Unless, I suppose, you believe in momentum.)
Another way to calculate the odds would be to use WAR and log5. That is how Greg Fertel has done it, and he does provide cross tabs. By his numbers, which are much more favorable to the Yankees (he has them winning the series almost 73% of the time), we will see a six- or seven-game series 58% of the time.
Think I chose the wrong reference class? Ok, let's try again. This time, we won't settle for a series shorter than seven games (anything less would be uncivilized). The odds of that happening? Greater than 27%.
Yes, the Yankees have proven themselves to be the stronger team over the course of the season. But does that preclude an exciting series? Not unless you make some pretty unjustifiable assumptions.
This is the virtue of MLB's playoff system compared with other sports (Bill James has even called for shorter playoffs in the NBA). The fact that we're guaranteed at least the possibility of an exciting series even between two teams of differing true talent is what makes it alternatively a luck-driven affair or good television. Except, you know, for the announcers.