SDMB Playoff History - Still a Coin Flip?

Frankly, Jeremy's recent snide blog comments almost dissuaded me from writing this post, but here goes anyway.

We now have 10 years of history in the books with four playoff teams, and it's reasonably clear that the first round of the playoffs is not coming out the same way that the championship round does.  First, the raw data (which obviously does not include the forthcoming Seattle-Maui series).

Cecil Cup

• When there's been a difference in records, the team with more regular season wins has won 7 of 19 series and 46 of the 103 games in those series. That's a series-winning percentage of .368 and a game-winning percentage of .446.
• When there's been a difference in records, the team with a better Pythagorean record has won 8 of 16 series and 46 of the 88 games in those series. That's a series-winning percentage of .500 and a game-winning percentage of .522.
• For the 19 series where I have game-by-game data, the home team has won 62 of 104 games. That's a winning percentage of .596.

Round One

• The team with more regular season wins has won 14 of 20 series and 67 of the 106 games in the Round One  series. That's a series-winning percentage of .700 and a game-winning percentage of .632.
• When there's been a difference in records, the team with a better Pythagorean record has won 10 of 18 series and 55 of the 97 games in those series. That's a series-winning percentage of .556 and a game-winning percentage of .567.
• The home team has won 51 of 106 games. That's a winning percentage of .481.

Here are the conclusions I've tentatively drawn.

1. A hundred or so games and 20 series is still a small sample size, so everything needs to be taken with a grain of salt.

2. It does appear that the "better" regular season team has an advantage in Round 1 beyond mere chance. I haven't looked at the statistical significance of the numbers, but that seems relatively clear.

3. I believe most of the variance in the results is explained by the larger gulfs between playoff teams. In the first eleven years of the league, when there was only one round of playoffs, the biggest gap in regular season records between championship contenders was eight games (in 2012). In the last ten years, in Round One we have seen regular-season-win gaps of 35 games, 29 games, 27 games, 26 games (twice), 23 games, and 21 games. Only once in those seemingly lopsided series has the team with fewer wins prevailed (2014, Cleveland over River Cities). If you take those seven series out of the data, and look only at Round One series where the teams were within 20 games of each other, the team with the better regular season record has won 8 of 13 Round One series. Those series appear more random and more like the Cecil Cup samples. I suspect that if I set the line at a 15-game gap, the numbers would get even closer to 50-50.

And so, I'm amending the much-derided "coin flip" theory.

I never really believed series were a true coin flip, as in a truly 50-50 proposition. The better team always had a slightly better chance of prevailing. I just thought that the true odds were far, far closer to 50-50 than most people believed. I thought that the "better" team only advanced maybe 55% of the time, which is close enough to a coin flip to make little difference.

Now I see that in series where there are large gaps in regular season records, it's not a coin flip. It's more like an NBA playoff series where the favorite usually, but not always, wins.

As always, I'm happy to respond to any questions or comments.