May 28, 2007

Anaheim's odds

Since the west never plays the east there is little information to go by in predicting the up coming series. I'm a firm believer that the west is better than the east on average, but this doesn't mean the best team in the east is necessarily worse than the best team in the west.

There are many mathematical techniques one can use to normalize team's opponents. Often people apply the simplest technique such as a first order strength of opposition calculation. I like to make things as complicated as possible so I've built a model that solves a complicated non-linear system of equations to determine a teams ability to score goals independent of opposition and ability to prevent goals independent of opposition. All the results are in a nifty little Google spread sheet. You can see on the bottom the west is better by a margin of about 3-4%, which makes a lot more sense than maybe the season series (west vs. east) might indicate.

Using these normalized goals, I went back and calculated the odds (based on a Pythagorean expectation (PCT)). From there, I calculated the odds of winning the series (S). Included in the table is the expected goals for and against in each series and the total goals per game. Take note this method predicted Ottawa beating Buffalo. Sample sizes is too small to have any real meaning and there should have been a lot more upsets if they percentages are correct, but from a theoretical perspective it's working quite well.

The main point: Anaheim's odds of winning the series is 60% once you've normalize out strength of opposition. Take note also 40% aren't bad odds, so Ottawa still has a decent chance of winning.

No comments: