Methodology Deep Dive

How CFBTrack handles incomplete seasons.

Incomplete seasons are handled as a data-quality problem, not a chance to fill gaps with invented certainty.

Open related toolBack to methodology hub

Plain-English explanation

When a current or future season is not populated enough for a page, the route should either explain the gap, fall back to the most recent usable season, or hide a module that cannot be supported.

The goal is to keep the page useful while making the provisional state visible to users and crawlers.

Inputs used

  • Available season lists from the data service.
  • Route-level filter state, selected season, and whether the requested sample contains rows.
  • Dataset-specific coverage notes such as betting-line availability, CFP proxy years, and player-stat history gaps.
  • Visibility, sitemap, and metadata rules for public routes whose content can change by coverage state.

What the model rewards

  • Pages that name the missing coverage instead of returning a blank or misleading result.
  • Fallbacks that use the newest complete season when a selected season is not usable yet.
  • Small-sample suppression where a model would otherwise look more precise than it is.

What the model does not claim

  • It does not claim a fallback season is the same thing as the requested season.
  • It does not claim missing records mean a team or player had no production.
  • It does not rank provisional data as if every source field has settled.

Example using Texas

If a Texas season is requested before every advanced stat, betting line, or player-stat row is loaded, a page should either show a visible caveat or move the user to the most recent complete context for that tool.

  • Scatter pages can recommend a fallback season when the requested season has no usable field.
  • Talent Usage can hide expectation lines when the filtered sample is too small.

Known limitations

  • Different data families refresh on different cadences.
  • Some historical fields are structurally thinner and may never match modern coverage.
  • Visible caveats reduce confusion but do not replace source-level corrections when the underlying record is wrong.