College Football DataCollege Football Data
  • CollegeFootballData.com
  • Get API Key
  • Legacy Swagger UI
  • Getting Started
  • Official libraries
  • Methodology and resources
  • API Reference
Helpful Links
  • CFBD
  • Patreon
  • Gumroad
Data + Code Packs
  • Starter Pack
  • Model Training Pack
  • AI Launchpad
  • AI Builder Pack
Other Resources
  • Rad Sports Analytics
  • Blog
  • Basketball

A Rad Sports Analytics platform.

xdiscordgithub
Methodology overviewMetrics and definitionsPredicted Points AddedWin probabilityWEPA and adjusted metricsElo ratingsSRS ratingsCORE ratings
powered by Zudoku
Methodology and resources

CORE ratings

Context & Opponent-Relative Efficiency (CORE) measures how much value a team creates on offense and allows on defense after accounting for game situation and opponent strength. It is designed to separate team performance from the contexts and schedules in which that performance occurred.

CORE is an efficiency rating, not a forecast, point spread, or percentage.

What goes into the rating

CORE uses qualifying pass and rush plays from completed FBS-versus-FBS games. Each play is first compared with other plays that occurred in similar football situations, including down, distance, field position, score, time remaining, home-field context, and play type. The results are then adjusted across the schedule so that offensive and defensive performance is evaluated relative to the opponents faced.

The model excludes overtime, garbage time, FBS-versus-FCS games, special teams, and non-scrimmage plays. Plays without the context required for a valid comparison are also excluded.

This is intentionally a high-level description. CORE is a proprietary CFBD model, so its fitted parameters and full calculation details are not published.

Reading the values

  • offense is points created above average per 100 qualifying offensive plays. Higher is better.
  • defense is points allowed above average per 100 qualifying defensive plays. Lower is better.
  • overall is offense minus defense. Higher is better.
  • offensePlays and defensePlays show the samples behind the two components.
  • throughSeasonType and throughWeek identify the inclusive endpoint of the snapshot.
  • modelVersion identifies the methodology used for the stored result.

The API returns the stored calculation precision. Round values only for display, and do not label them as percentages.

Coverage and limitations

Public retrospective CORE ratings begin with the 2016 season. Historical ratings are retrospective results from the released methodology rather than a record of what the model would have reported at that point in time.

CORE describes team offense and defense over the qualifying plays in a stored snapshot. It does not include special teams, player ratings, matchup projections, spreads, or win probabilities. Early-season results also carry more uncertainty because they are based on fewer plays and a less connected schedule.

Use the CORE ratings operation to filter by season, team, or conference. A season or team is required.

Edit this page on GitHub
Last modified on August 8, 2026
SRS ratings
On this page
  • What goes into the rating
  • Reading the values
  • Coverage and limitations