Welcome

Welcome to the Elo Sports blog. I'm Nick, and I'll be taking you on a magical tour through NCAA basketball statistics and history in an attempt to accurately rate teams for the 2013 NCAA tournament. As with any statistical rating algorithm ours will not be perfect. But that's the point of this whole process: to iterate, refine and test different rating systems and algorithms until we have one we feel comfortable using this March.

The first rating system we'll implement was developed by Árpád Élő, a Hungarian Physicist and avid chess player. The Elo system was originally designed as an improved rating system for chess but has since been used to rate the relative skill level of players in many other games and sports as well. While the rating system was designed for a two-player game, it's possible to apply it to team sports simply by considering each team as an individual "player". The athletes or competitors that make up each team can be considered different strengths and weaknesses of the "player". Just as certain chess players might excel in particular areas of the game, so too do different basketball team excel in particular areas depending on the roster and coaching. Finally, we'll consider roster and coaching adjustments between seasons as a player improving or weakening over time in the two-player game, something we'd want the rating system to be able to take into account.

The Elo system at a glance:

• Each team is assigned a numerical rating.
• In each game the statistically likely outcome is calculated. Each team is given a probability of winning.
• Based on the probability of winning and the actual result a new rating is calculated for each team
• A team with a low probability of winning a game will see their rating increase by more if they win, while losing will result in a more modest decrease. The converse is true for teams with a high probability of winning.
• Ratings are recalculated after every game throughout a season. The more games you have the more accurate the ratings become.

The first step in this process will be modifying the Elo system to suit our particular needs. If we can build a system that we're comfortable with, applying it to different sports will be the next step. In the next several posts I'll go over several of the more interesting modifications we'll need to make to the pure 2-player game that the Elo algorithm was originally meant to rate. These include things like the home-field/court advantage, season-to-season rating resets, and stronger vs weaker conferences and schedules. So stay tuned for some riveting sports and math discussions!

With that, I'd like to finish by saying thanks for reading, and leave you all with a great Lonely Island video that should fairly accurately describe the tone and direction this blog will take in the coming months.