The Signal WARP ratings are based off of play-by-play data. Every single non-garbage play from REGULAR_PLAYS, PUNT_PLAYS, and KICK_PLAYS is considered holistically to form a rating for each team. It involves, in part, some LINEAR ALGEBRA to generate the final numbers. However, the weighting of plays is derived from more advanced statistics. POINTS_PREDICTIVE, or ABOVE-AVERAGE POINTS, represents a margin of victory (or defeat) against a hypothetical AVERAGE team. If team A has a rating of 27.43, I would expect them to defeat an AVERAGE FBS team by a MARGIN of 27.43 points. POLL_POINTS, or RESUME, represents a better measure of RETRODICTIVE team measurement. It is based off of how well a team performs relative to the average team, but having more weight on the act of WINNING games. This should not be used as a predictive measurement. This system does not use talent priors or returning production, unlike many other systems such as SP+ or ESPN'S FPI. That means it will be slightly worse toward the beginning of the year. The only priors used are performance from the last two seasons. I have never played a down of football in my life. I got interested in football thanks to the OHIO STATE UNIVERSITY MARCHING BAND and the FIGHTIN' TEXAS AGGIE BAND. I have seen the OSUMB in-person once. A brief list of teams I support: NCAA D-I A: - Texas A & M - Ohio State - California-Berkeley NCAA D-I AA: - Stephen F. Austin - California-Davis NFL: - Pittsburgh Steelers NBA: - Portland Trail Blazers MLB: - Anaheim Angels NHL: - Anaheim Ducks Thank you to WESLEY N. COLLEY and KENNETH MASSEY for inspiring my interest in sports ranking systems both retro- and pre-dictive.