MLS Playoff Math: Using Statistics to Predict the MLS Cup Champion

MLS CUP MATH

BY BILL REESE
Contributor

All season long I’ve been building models to predict the results of MLS games using a formula that involves each team’s goals-per-game average. It is by no means perfect, but it does serve as an indicator of how well clubs play at home and on the road.

Back on August 5, when I debuted the goals/game predictor on Google Drive, the program predicted that New York would win the shield on the first tiebreaker with Portland. As New York dropped points at Columbus and Chivas, they gradually fell from—then regained—that top spot. At times this season, the predictor thought Seattle, Portland, Montreal and L.A. were mathematical favorites to win the Shield. New York’s performance down the stretch earned them their first title, but no club in recent memory has benefited so much from the missteps of so many others.

Now, I know math is held in a slightly higher regard than DC fans in these parts, but bear with me while I explain how it works.

The formula has 3 components:

Home Goals-Per-Game (HG/G): Very simple, goals scored at home divided by matches played.
Road Goals-Per-Game (RG/G): Needs no explaining
Defensive Multiplier (DM): This is my creation. It takes a team’s goals allowed-per-game at home and on the road and compares them to the league average. This season, the MLS average for goals allowed at home was 1.04 G/G. On the road it was 1.58. This figure is then multiplied by their opponent’s goals/game average. The more any club’s goals allowed ratio deviates from the league-wide norm, the more their opponents goals scored average is affected.

Thrilling stuff, right? I know you only care about who’s going to win the games, so let’s get to the heart of the matter, starting with the Eastern and Western Conference Play-In Games:

Pages: 1 2 3 4 5 6

  • Dave from Dix Hills

    So ……. Much …….. Math …… Wouldn’t hurt my brain as much if Red Bulls won at the end though.

  • Aguinaga

    Fun exercise. Only thing missing from the analysis is the statistical probablity that the predictions made all come to pass :)

    • Bill Reese

      Well, there can only be 3 outcomes to a match (not including penalty kick wins, which go in the book as draws, technically). So, I have a 33% chance of making a correct pick. There are 15 playoff games. So you’d go .333 to the 15th power, which is a number I cannot begin to comprehend. My best calculation is that you would have a 0.00167% chance of picking all 15 correct in a world where Win, Loss and Draw had even chances of occurring. That translates to a 1.67 chance out of 1,000