A Node routine to take data and rank college football teams. The overall goal here is to produce a ranking that mimics the College Football Playoff committee. So far, it has been pretty close, usually getting the top 4 to 6 teams correct with reasonable rankings for the rest of the top 25. The most notable difference from a typical ranking is that big wins or losses take effect immediately. For example, ASU rose much farther than most other polls after beating ND because there is no stickiness to hold them down or prevent teams above them from getting passed.
The data is pulled from http://prwolfe.bol.ucla.edu/cfootball/scores.htm.
Once the data is downloaded, teams are ranked by number of losses with opponent's season score differential per game (excluding the current team) as the tiebreaker.
From there, it attempts to rearrange the teams to minimize the number of upsets that have occurred. For example, if a team has 1 loss to a very low ranked team, that team will be moved down until it reaches rough equilibrium.
Once that ranking is fairly balanced, it takes a weighted average between that rank and the initial loss/differential rank to produce the final value.