Cubs Win World Series in 7; Playoff Prediction Methods Explained

Major League Baseball Playoff Prediction

I'd like to shed some light on how Baseball Playoffs Now predicts the postseason. Our final record predictions generate division placements and the wild card team, based on a history-weighted pythagorean method. Once we know the playoff teams, we turn to a completely different statistic to figure out who are the better teams.

Our Power Rankings are compiled using both pure wins/losses and runs scored/allowed in two separate algorithms, both rated against an infinite strength of schedule matrix. But run differential is the best single predictor of upcoming performance, so Baseball Playoffs Now weights that factor much more heavily when creating a special stat called Playoff Ratings. The only use for this statistic is picking winners in our playoff predictions; we don't publish it as Power Rankings because it is skewed too far away from wins and losses.

Using Eric Seidman's Home Field Advantage article to determine HFA, we give a (small) home field advantage to the proper teams' ratings. The winner of the series is therefore simply the team with the higher rating.

Mathematics, backed up by Seidman's data, also returns the probabilities that a series will extend to 3, 4, or 5 games (in the case of the Divisional Series) or 4, 5, 6, or 7 games (in the case of the League Championship and World Series):

The math: (coin flip for winner [or use historical data, but there is only a very small difference] ^ number of games in series) x number of possible outcomes in which the series lasts that long

Divisional Series:
3 games = 0.5^3 x 2 = 25%
4 games = 0.5^4 x 4 = 25%
5 games = 0.5^5 x 16 = 50%

League/World Series:
4 games = 0.5^4 x 2 = 12.5%
5 games = 0.5^5 x 8 = 25%
6 games = 0.5^6 x 20 = 31.25%
7 games = 0.5^7 x 40 = 31.25%

Using these numbers, we use some fuzzy math to generate the least-scientific prediction of this entire site: series lengths. We figure out the greatest difference in Playoff Rating between teams in the playoffs, then figure out about where the two teams in each series lie in that range. Plugging that percentage into the probabilities above, we can approximate series lengths.

And voilà!

American League Divisional Series
No. 1 seed CHW over no. 4 seed TAM in 5 games.
No. 2 seed BOS over no. 3 seed OAK in 5 games.

National League Divisional Series
No. 1 seed CHC over no. 3 seed ARI in 3 games.
No. 2 seed PHI over no. 4 seed STL in 4 games.

American League Championship Series
No. 1 seed CHW over no. 2 seed BOS in 6 games.

National League Championship Series
No. 1 seed CHC over no. 2 seed PHI in 5 games.

*** World Series ***
*** No. 1 seed CHC over no. 1 seed CHW in 7 games. ***