Alpe d'Huez said:I just can't see the Royals beating the Orioles. Then again, I thought the Angels would beat them in 4 games without much of a problem.
Amazingly, FanGraphs not only favors the Royals, but has their win probability for this series as 63%. That is a very large margin for baseball.
Their basic argument is that some of the Royals hitters, like Butler and Moustakas, underperformed during the regular season, and are better than their regular season stats indicate. This is the tricky part of using stats. There is a natural tendency to weight heavily recent performance, whereas a lot of studies show that you have to go back several years (when the data are available) to gauge the value of players. With the two aforementioned players, the argument is strengthened by the fact that their BABIP--% of balls put into play that go for hits--was much lower this season than their historical average, which suggests some bad luck.
In the saber view, stats are like coin flips. It doesn't matter if the last five tosses have been all heads or all tails, the probability of the next one remains 50/50. The weighting of the coin is determined by several years of data, and once that has been determined, the odds are unaffected by anything that has happened in the past few weeks, or even very much by the past few months.
The larger point, and my pet peeve, is that the playoff system is woefully inadequate for determining a seasonal champion. Chance plays far too large a role in a short series. The best team--as determined by the long haul of the season, and not by wins, really, but by run differential, or even better, by weighted on-base average differential--frequently loses out. If one team is gauged to be a 55 vs. 45 favorite to win a single game, based on the strength of its seasonal record, it still has only about a 60% chance of winning a best of seven series.
We've talked in the NFL thread about the problem of expanded playoffs, and teams with losing records having a shot to win the SB, so this is a problem with all sports. But it's especially a problem with baseball, which requires a very large sample size to even out most of the random variation.