Methodology Deep Dive

How CFBTrack calculates Talent Usage.

Talent Usage asks whether a team or coach got more wins and development from a talent base than the comparison group would normally expect.

Open related toolBack to methodology hub

Plain-English explanation

The page treats recruiting as the input, then compares on-field results and NFL draft development against that input.

Expected lines are calculated from the filtered comparison group, so the answer changes when the user changes seasons, entity type, result metric, or normalization mode.

Profile trend expectations are fit on season-level rows, while the main board expectation is fit on the aggregated entity rows shown in the table.

Inputs used

  • Recruiting capital from weighted roster value, rolling recruiting average, or blue-chip ratio.
  • On-field result score from final AP rank, latest available CFP committee rank, win percentage, or adjusted rating.
  • Draft capital from weighted NFL draft value or raw draft-pick count.
  • Selected seasons, entity type, conference filters, normalization mode, and combined-score weight preset.

What the model rewards

  • Teams and coaches that win more than their talent profile suggests.
  • Programs that turn recruiting input into more NFL draft value than expected.
  • Profiles that beat expectation across both winning and development when the combined preset is balanced.

What the model does not claim

  • It does not claim recruiting rankings are a perfect measure of player quality.
  • It does not claim a coach personally developed every draft pick attached to a roster window.
  • It does not show expectation lines when the filtered sample is too small.

Example using Georgia

For Georgia, a high recruiting baseline sets a high expected result. The page rewards seasons or windows that still outperform that expectation in wins, adjusted ratings, or draft output.

  • A loaded roster can post strong raw results but only modest win-over-expected if the model already expected elite results.
  • A less talented roster can score well when it beats the expected line by a wide margin.

Known limitations

  • Transfers and imperfect roster alignment can blur which talent belongs to a specific result window.
  • CFP mode uses the latest loaded Playoff Committee ranking for each season; seasons without committee ranking rows fall back to the top-four proxy.
  • Small samples suppress expectation lines instead of inventing precision.