How We Rank College Football Teams: A 9-Factor Power Score

Most college football ranking systems are black boxes. The AP Poll is a vibes check by media members. The CFP Committee meets behind closed doors. Even advanced systems like SP+ and FPI publish a single number without letting you interrogate what drove it. Our power rankings tool is built on a different premise: every input should be visible, every weight should be adjustable, and the methodology should be transparent enough that you can disagree with it intelligently.

The 9 Factors

Every FBS team receives a composite power score between 0 and 100. That score is a weighted blend of nine factors, each normalized to a 0-1 scale before weighting. Here is what goes into it.

1. SP+ Rating (Default Weight: 28%)

SP+ is Bill Connelly's efficiency-based rating system that measures what a team does on a per-play basis, adjusted for opponent quality and garbage time. It remains the single most predictive public metric in college football, which is why it carries the largest default weight. We normalize each team's SP+ rating relative to the full FBS distribution for the current season.

2. Win Record (12%)

Three-year win percentage, adjusted for actual strength of schedule. SOS is computed from opponent win percentages across the same window, not from preseason projections. A 10-2 record against a schedule averaging .550 opponent win percentage is worth more than 10-2 against a .420 schedule.

3. Recruiting (8%)

Composite recruiting scores from the 247Sports Composite, averaged over the most recent four classes. Talent matters: teams in the top 10 of recruiting rankings account for a disproportionate share of national championship appearances. But talent is not destiny, which is why this factor sits at 8%, not 20%.

4. Coaching (4%)

Head coach career win percentage, weighted toward recent seasons. A proven coach at a program with modest talent can still outperform expectations. We also factor in tenure stability, since coaching changes typically cause a one-to-two year regression.

5. Offense (9%)

Points per game, yards per play, and third-down conversion rate, blended into a single offensive efficiency metric. This captures both explosiveness and consistency, not just raw volume that can be inflated by pace.

6. Defense (9%)

The mirror of the offensive metric: opponent points per game, yards per play allowed, and third-down defense. Lower is better. Elite defenses that hold opponents below 18 points per game score near the top of this factor.

7. Turnover Margin (8%)

Net takeaways minus giveaways, not returning production percentage. After running an ANOVA analysis on factor contributions to predictive accuracy, we found that actual turnover margin explained more variance in game outcomes than several other candidates. Teams that consistently win the turnover battle tend to outperform their other metrics.

8. Portal Impact (12%)

The transfer portal has fundamentally changed college football roster construction. This factor is a 50/50 blend of net portal impact (quality of incoming transfers minus departures) and experience-weighted roster depth, with 85% of experience attribution coming from name-matched NCAA years of eligibility. Programs that reload effectively through the portal, like USC and Colorado in recent cycles, get a meaningful boost here.

9. Momentum (10%)

A forward-looking signal: 60% year-over-year SP+ improvement plus 40% portal-era wins above expected (post-2022). Teams on an upward trajectory, improving efficiency and exceeding their talent level, carry momentum that often persists into the following season.

How Weighting Works

The default weights above reflect our analytics-optimized preset, tuned by analyzing which factor distributions best predicted actual game outcomes in held-out seasons. But the Rankings tab gives you nine sliders to set any weights you want.

This is not a gimmick. Different philosophies about what makes a football team "good" are genuinely defensible. If you believe recruiting is king, crank that slider up and watch blue bloods rise. If you care about on-field results above all else, the Traditional preset rewards teams that actually won games. The point is that you can see exactly what changed and why.

What Makes This Different from AP/CFP Rankings

Three things separate a transparent composite system from the rankings most fans follow.

Customizable weights. No other public ranking tool lets you adjust the formula in real time and instantly see how it reshapes the top 25. The AP Poll gives you one list from 62 voters. We give you a system you can interrogate.

Transparent methodology. Every factor is explained, every data source is public. We use the CFBD API for game data, the 247Sports Composite for recruiting, and Bill Connelly's SP+ for efficiency ratings. Nothing is proprietary or hidden.

Predictive validation. We backtest the model against actual game outcomes. On 2024-2025 holdout data (games the model never saw during training), the analytics preset correctly predicted the winner 83.6% of the time. That is competitive with published models from ESPN (FPI) and Pro Football Focus, but ours is the only one you can take apart and rebuild.

Predictions and Head-to-Head Matchups

The same engine that powers the rankings also drives our game prediction tool. When you select two teams, the model computes their composite scores, then applies three matchup-specific adjustments: home field advantage (historically worth about 3 points), rivalry intensity bonuses drawn from our 15-rivalry database, and momentum differential between the two programs.

The result is a win probability and a predicted margin. You can also explore how individual factors compare between any two teams using the comparison tool, which renders tug-of-war visualizations for each of the nine factors.

The Data Pipeline

All data is pre-baked into static files, meaning the app makes zero API calls at runtime. A separate data pipeline pulls from the CFBD API, ESPN's public scoreboard, and the 247Sports Composite, then writes compressed .rds files that the app loads at startup. This keeps the app fast (sub-second load times) and independent of external API availability.

The dataset covers 137 FBS teams (including transitioning programs like North Dakota State joining the Mountain West in 2026), approximately 9,800 games from 2016 to present, 8,800 recruiting entries, and 796 coach-seasons.

Try It Yourself

The best way to understand the system is to use it. Open the Rankings tab, pick a preset, drag a few sliders, and watch how the top 25 reshuffles. Then pick two teams and run a prediction. The goal is not to tell you what to think about college football. It is to give you better tools for thinking about it yourself.