Here is the full post:
have developed a 100% mathematical (and therefore theoretically unbiased) power ranking system that I will be using to rank our Conference on a game-by-game basis. The official pre-season rankings won't be released until tonight after training and to hopefully include as many offseason boosts as possible, for optimal accuracy. In the meantime though, I want to use this post to fully disclose the formula. If you don't care, that's fair, this will definitely be tl;dr, so just enjoy the rankings when they come out. For the mathematically inclined though, enjoy:
The system has two components. One component is a power ranking based on the current starting lineup which I will explain first. The preseason rankings will be based entirely on this factor. The second component will an adjustment based on actual game results, that should hopefully, over time, account for errors in the roster rankings.
Roster Power Rankings: To begin with, I put every team's starting lineup in a spreadsheet. Right now, I'm looking ONLY at the level of each starter. Backups and builds are ignored (this is the primary source of error that component two will have to adjust for). CPU players are universally treated as the equivalent of a level 1 human player, regardless of the CPU's actual level. Holes in the lineup are treated as level 0. Each player is then graded against the other players at his position (WR1 and WR2 are considered separate positions, as are LOT and ROT, etc...). A player who is the highest level in the Conference at his position will be rated 100, anyone else will be linearly scaled (so if the best player were a 10, an 8 would become an 80, etc...).
Next, using the scaled player ratings, I give each team a separate rating in pass offense, rush offense, pass defense, and rush defense, using weighted averages of the players' relative importance to that aspect of the game. These are the crude formulas I'm using:
Pass Offense: (QB + (WR1 + WR2)/2 + TE/5 + (OL AVG)/2)/2.7
Rush Offense: (HB + FB/5 + (OL AVG))/2.2
Pass Defense: (CB1 + CB2 + (S AVG) + (DE AVG)*.75 + (LB AVG)/2)/4.25
Rush Defense: ((DT AVG) + (DE AVG)/2 + (LB AVG))/2.5
Then, I average the pass and run numbers together to get overall Offense and Defense rankings, and finally I average these numbers together to give the team its final Overall Power Ranking.
Adjustments: The power ranking is obviously not perfect. It will change each week as rosters change and players level up, but it also has systematic flaws that will get exposed when teams are "upset" by lower rated teams. To adjust for these upsets, any team that defeats a higher rated team (based on roster ratings AT GAME TIME) will receive an "upset bonus" equal to the difference in roster ratings. The higher rated team will receive an equal "upset penalty". These bonuses and penalties will be capped at 5-10, I'm not sure yet. Teams that are overrated in the roster ratings (because they lack depth, or their starters are poorly built and perform below their level, or because of poor game planning and tactics, or whatever other reason) will see their ranking drop, even as their roster rating remains high, as they lose games they "shouldn't" have. Underrated teams, with positive intangibles, will similarly see their ranking rise.