40%
Actual roster composition
Position-level quality, depth, college production, returning production and transfer fit.
Capital Rank v1
Method before outcome.
A transparent editorial synthesis of what a program assembled before kickoff—not a claim of false mathematical precision.
Conceptual weights · ordinal result
The four-part framework
What drives Capital Rank
The weights impose research discipline, but the output is a rank—not an invented decimal score. Source quality, uncertainty and fit shape the final editorial judgment within each category.
30%
Head coach & staff
Program leadership, coordinator quality, development record, scheme fit and staff continuity.
15%
Estimated roster spending
Reported resource bands that combine revenue sharing and NIL or collective capacity.
15%
Continuity & integration risk
Returning roles, line cohesion, system change, portal volume and the difficulty of assembling new parts.
A critical distinction
Pedigree is potential. Production is evidence.
Useful for measuring physical upside, depth and the quality of the talent base entering college.
Useful for measuring whether players have already translated opportunity into effective football.
Capital Rank considers both. A developmental roster can outrank its original recruiting labels; an unproven blue-chip collection does not receive automatic certainty.
Spending has limits
The 2026 bands are reported estimates from an industry survey. They combine revenue sharing and NIL or collective resources, and are neither audited payroll nor a promise that every dollar was deployed efficiently.
Confidence is separate
Research confidence describes assessment stability, not the team’s ceiling or a wager. Integration risk, uncertain roles and staff change can lower confidence even when the roster grade is strong.
Unranked is not rejected
All 68 programs now have researched dossiers. The 43 outside the frozen Top 25 remain explicitly unranked—there are no hidden ranks 26–68.
Forecast Book v0.1
Blind first. Market second.
The experimental Forecast Book converts roster grades, frozen rank tiers, quarterback, skill, trenches, staff, continuity, known availability and venue into a half-point fair margin. The same rules apply to every game.
Each payload is created before kickoff, canonicalized, hashed and committed before the market is retrieved. Its displayed range is editorial uncertainty—not a calibrated confidence interval—and roster-research confidence remains separate from betting confidence.
v0.2 decision audit · research only
Open the prospective validation book The current site now separates model direction, the frozen official position and the exact v0.1 restriction. Near-pick'em forecasts are not assumed to erase against-the-spread direction, but no cover probability, expected value or replacement classification is published without chronological calibration.
The candidate framework will preserve exact integer-margin mass, including pushes at key numbers, and will keep matchup input reliability separate from the forecasting system's overall experimental maturity. It is not active and cannot alter the frozen Week 1 record.
Future Live Power
A calibrated transition to the field
The preseason rank acts as a prior. As qualifying play volume accumulates, opponent-adjusted field evidence earns more weight and the prior earns less. That decline will be calibrated against history, rather than forced into an arbitrary week-by-week schedule.
Major injuries, suspensions, quarterback changes and meaningful usage shifts can create a separate availability adjustment. Closing market lines may later be stored as a benchmark, but are not an initial model input.
Explore the future architecture