SO you've heard the complaints this year about the playoff system going off the rails because many of the teams with more wins than their post-season competitors wound up losing (raise your hands as we call your names: Braves, Orioles, Dodgers, Rays).
And you've heard the countervailing comments that have been in force for many years before the further "wilding" of the Wild Card team had produced a general feeling of chaos matching what many feel about the world at large. Namely: "the playoffs are a crapshoot." (Joe Sheehan, crusty Baseball Prospectus renegade, is especially vehement--and numbingly repetitive--with respect to this mantra.)
So--as is usually the case--here we are, in Atlas mode, carrying the worldly weight of such tasks squarely on our shoulders...and providing some answers never seen before. Let's get to it...
FIRST, note the summary total (for all post-season games of all types, since the invention of the World Series in 1903). It shows that teams with better records than their opponents (column marked "B") only hold a slightly better than 50% success rate. That percentage was higher in the "pennant era" (1903-68) and the first era of division play, which we call "the championship series era" (1969-93), but has dropped since.
Note that since 2000 this percentage is less than 50% (all such occurrences on this distribution table are shaded in blue).
WE then go on to break down the World Series by length of games. Note that as the number of games in the World Series increases, the likelihood of the team with the better record being the winner declines (from 65.5% in sweeps to just 43.3% in seven-gamers). But note also that this trend is reversing itself for seven-game series since 1995, which kicked off what we call "the wild card era."
SWEEPS are where the better teams do best at holding their own (with one exception, that we'll look at it later). The World Series sweeps (4/0), now somewhat mixed in with ALCS/NLCS data, and the old CS format and Division series sweeps (3/0) yield an aggregate success rate for 71% for teams with the better won-loss record.
When we get into the Division Series' 3/1 and 3/2 subgroups, we see a stark decline: the 3/1 series are only 50/50 for the better team, and the 3/2 series, fueled dramatically by the changed dynamics in the "wild card" era, have been on an excoriating run against the better teams, who've won barely one-fourth of those series since 1995. This is the hidden story of how "the playoffs [became] a crapshoot."
And similarly, in the best-of-3 series (the variants of the more recent "wild card era" approach), the "sweeps" (the 2/0 series results) have thus far pushed against the team with the better record. In the old "one and done" variant of this approach, you can see that the results when broken out in the three categories--better (B), the same (E), and worse (W)--are virtually random.
All of these shorter series are currently running against better teams, who've won only forty percent of the "one and done" and "best of three" series.
SO this is the structural anatomy of the post-season, and how it's changed as the post-season itself has changed. How this might look in twenty years absent any better system (and readers here know by now that we have several of those, including a new one that we'll discuss here after the World Series) is really anyone's guess, but there's no reason not to think that it will remain reasonably similar to what we see here.
It can be argued that we should break these categories (B/W) into smaller units that show the range of difference; while the sample would be smaller, it might provide some additional nuance. It can also be argued that we should use Pythagorean Winning Percentage for this breakout--and if that data were readily available, we'd do so...but none of the reference sites has seen fit to compile this in an easily accessible framework, and until they do so, we are constrained by time issues from doing so. (If someone provides it to us, however...)
