There are two areas of basic methodology to cover: Ratings and Predictions.
Ratings Methodology
One of the things that makes sports so exciting is the debate that inevitably ensues after every week of games about which team or group of teams are better than the others. The human polls tend to be weighed down by emotion and bias, and the computer polls are limited by measurable data (they can’t actually watch the games). Our goal was to make a set of ratings that use common sense by trying to mimic human polls with a solid statistical element driving it.
Statistical theory is very important, but experience has taught us that theory rarely translates to practicality. In general, the underlying basis of our ratings are built around logistic regression, and more specifically, the Bradley-Terry model. One can learn more about Logistic Regression in this Wikipedia article. This method is, in fact, not much different than a lot of ranking systems already in place. Our system differs in that it weighs each game differently based on a variety of factors that we humans use either consciously or subconsciously to rank teams.
Predictions Methodology
When DRatings was first built, we were hyper-focused on rating teams. Eventually, it became apparent that it was easy to convert our ratings into predictions. Each sport is treated very differently and to varying degrees of accuracy. As a starting point, we use the team ratings to formulate a base prediction.
Let’s look at an example as it applies to two teams:
We want to know who is going to win the 2012 SEC Championship which has been set between National Championship hopefuls Georgia and Alabama.
After all of their regular season games in 2012, Alabama has a rating (x1) of 3.2712 and Georgia has a rating (x2) of 2.7662.
The formula for finding the probability of winning p(x1) is exp(x1-x2)/(1+exp(x1-x2)).
That means that the probability of Alabama winning the SEC Championship was 62.4%.
But there are so many other factors that might influence the outcome of a game: injuries, starting pitcher, weather, etc. We have added varying degrees of adjustments to different sports. Keep an eye out for these adjustments as we mention them on the sport’s prediction pages.