Conference Power Rankings: What the Scoring Data Actually Says

Analysis · by Rick's Picks Analytics

Forget the eye test and preseason media polls. We built conference power rankings from raw scoring data, head-to-head records, and postseason performance across a decade of college football.

Per-Conference Scoring Profiles

For every conference in every season from 2015-2024, we compute three metrics:

  1. Average Points Scored -- offensive output
  2. Average Points Allowed -- defensive quality
  3. Scoring Differential -- the gap between offense and defense

We compute these from every game appearance (not just conference games), properly weighting so each team's games count equally regardless of whether they're home or away.

Why proper weighting matters: A naive approach averages home-game scores and away-game scores separately, then combines them. But teams play different numbers of home and away games, and strong teams might play more home games in a given season. Our method counts each game appearance once from each team's perspective, avoiding double-counting and weighting biases.

The conference hierarchy: The resulting rankings show clear tiers. Power conferences (SEC, Big Ten, Big 12, ACC) consistently outscore Group-of-5 conferences in both raw scoring and scoring differential. Within the Power tier, the SEC typically leads in scoring differential (strong offense AND strong defense), while the Big 12 leads in raw scoring (strong offense, weaker defense).

Power vs Group-of-5 Head-to-Head

The question: In games where a Power-conference team faces a G5 team, how often does the Power team win?

The test: Binomial test against 50%, with Cohen's d on the scoring margin. This uses our season-aware conference mapping, so teams are correctly classified even across realignment years.

What the data shows: Power teams win these matchups well above 50%. The talent, resource, and depth gaps are real and measurable. This is not a controversial finding, but the specific win rate and margin help calibrate our cross-conference predictions.

The spread angle: Vegas already accounts for the Power-G5 gap, so the straight-up win rate doesn't directly translate to ATS value. The ATS question is whether Vegas prices the gap correctly. Our data helps answer that.

SEC vs Other Power Conferences

A deeper cut: Within Power-on-Power matchups, does the SEC outperform?

The test: Binomial test on SEC win rate vs other Power conferences. We fixed a bug in the original code that was filtering on wrong DataFrame columns, which corrupted the sample.

What the data shows: The SEC does win Power-on-Power matchups at a rate above 50%, confirming the finding from our conference hypotheses study. The effect replicates on the 2023-2024 test set, giving us confidence it's a real structural advantage rather than a statistical artifact.

Bowl and Playoff Performance

The postseason question: Which conferences perform best in bowl games and playoff matchups?

We filter for games in weeks 14-17 (conference championships and postseason) and compute win rates by conference.

What the data shows: SEC and Big Ten teams have performed well in postseason play, while some conferences have struggled. However, bowl game sample sizes are small for any single conference in any single season, making it difficult to achieve statistical significance.

The selection bias: Bowl matchups are not random. The committee and bowl selection process creates pairings that may not reflect a fair cross-section. A conference that sends its 3rd-best team to a bowl against another conference's 6th-best team will look stronger, even if the conferences are comparable overall.

How Conference Analysis Informs Predictions

Conference metrics serve as contextual modifiers in our prediction system:

  1. Cross-conference calibration: When two teams from different conferences meet, conference-level scoring data helps calibrate the expected scoring environment.
  2. Conference strength modifier: A 7-2 record in the SEC means something different than 7-2 in the Sun Belt. Conference strength adjusts our quality estimates.
  3. Bowl game projections: For postseason matchups between unfamiliar opponents, conference-level metrics fill gaps in head-to-head data.

These are background adjustments, not headline features. The primary prediction comes from team-specific ELO and game-specific factors (weather, travel, stadium). Conference identity refines the estimate at the margins.

Analysis uses binomial tests, Cohen's d, and Bonferroni correction. Conference assignments are season-aware to account for realignment. Power vs G5 classification uses the current era definition (Power 4 since 2024). Results validated on 2023-2024 holdout data.

Rick's Picks publishes statistical analysis of college football for informational and entertainment purposes. Nothing here is betting advice. 21+. If gambling is affecting you or someone you know, call 1-800-GAMBLER.