CFB Power Rankings — Preseason —
| # | Team | Rating | Trend | Our Take | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Loading rankings… | ||||||||||
Methodology — how we build the rankings
The Signal Labs CFB Composite Rating combines two layers: a consensus anchor (median of five external power-ranking sources) and a six-component model. The blend weight between them shifts by season phase.
Phase-aware blend. Preseason, we have zero on-field data — so the consensus median dominates (75% anchor / 25% model). Each week of play earns the model more weight, until by the playoff stretch the model dominates 95/5. Preseason: heavily consensus-anchored — no current-season data yet. In-season: our model takes over progressively as games are played.
| Phase | Consensus Weight | Model Weight |
|---|---|---|
| Preseason | 75% | 25% |
| Weeks 1–3 | 60% | 40% |
| Weeks 4–7 | 40% | 60% |
| Weeks 8–11 | 20% | 80% |
| Weeks 12+ | 5% | 95% |
Model components (each z-scored vs the 136-team FBS field, projected onto a 1500-mean / 50-sd scale):
1. Adjusted Strength (40%) — Current opponent-adjusted strength rating.
2. Returning Starters (20%) — Offensive + defensive returning production %.
3. Transfer Portal Net (10%) — Incoming star-rating sum minus outgoing.
4. Coaching Continuity (10%) — Penalty for HC / OC / DC changes.
5. Prior Season (12%) — Last full season's adjusted rating.
6. Recruiting Class (8%) — Incoming-class composite ranking.
Tier bands are fixed by rank: Playoff Contender (1–10), NY6 Contender (11–25), Bowl Eligible (26–50), Building (51–100), Rebuilding (101+).
What this is not: a prediction. In preseason, we look a lot like the consensus median — that's intentional. The story is the divergence arc: as the season progresses, our model picks up signal that consensus polls smooth over for narrative reasons (opponent-adjusted in-season ratings, drive efficiency, pace splits, weather). Watch the "vs Consensus" column widen by week 6.