The Wire Power Rating has four components. I'll walk through each one so you know exactly what's going into the number next to each team's name.

Component 1: The preseason rating

Every team starts with a preseason number. That number has two inputs:

  • Prior year baseline (60%) — what we knew about the team coming out of last season
  • Roster grade (40%) — who's returning, who's new, and whether the coaching is stable

The roster grade itself has three buckets: returning production (50%), incoming talent (35%), and a coaching modifier (15%). Within each, there are specific sub-weights. A returning starting quarterback is worth more than a returning running back. A transfer portal addition is worth 1.3x a freshman recruit with the same composite rating. Coaching overperformance gets more credit than underperformance gets penalty.

Component 2: The information score

This is what makes the Wire different. Every week, each team earns "information points" based on who they played and how surprising the result was. A top-10 team blowing out another top-10 team earns maximum info. A top-10 team beating an FCS opponent by 40 earns almost nothing.

As information points accumulate, the preseason rating gets replaced by actual on-field performance. By the time a team has played around 30 information points' worth of meaningful games, the preseason prior is gone. The rating is 100% driven by what they've done this year.

This is why the Wire will diverge sharply from other models by late October. Most systems are still anchoring to preseason expectations. The Wire isn't.

Component 3: The performance rating

The in-season engine measures team strength across offense, defense, and special teams. The inputs are:

  • EPA per play — the single most predictive metric in football analytics
  • Success rate — the consistency measure
  • Explosiveness — big-play ability
  • Finishing drives — red zone execution
  • Field position — mostly a special teams signal
  • Opponent-adjusted margin of victory — capped at 28 points to prevent blowouts from distorting things

All metrics get iteratively adjusted for opponent strength. A 40-point win over a good defense counts way more than a 40-point win over a bad one.

Component 4: Game-day adjustments

When we predict a specific game, we take both teams' ratings and layer on situational factors:

  • Home field — tiered by venue quality (4 points at Death Valley, 1 point at a half-empty stadium)
  • Rest — bye weeks and short weeks both count
  • Travel — matters for cross-country trips
  • Game type — rivalries shrink spreads, bowl games are neutral sites

Simple subtraction of two team ratings, plus these adjustments, gives you a predicted spread for any matchup.

What's next

The Wire Poll gets built on top of the WPR by adding a W/L resume layer — quality wins add points, bad losses subtract them. That's what makes the two diverge and what makes the comparison interesting.

If any of this is unclear, the beauty of running this publicly is you can tell me. Every methodology tweak will be documented here.