Model details
Title-game appearance model · one probability for each team.
Each of 50,000 simulated seasons follows the conference tiebreakers in order and awards the qualifying spots. The displayed chance is the share of seasons in which your team qualifies. Fixed seed: 2026.
Missing SportSource ratings and future CFP ordering use full national FBS results, simulated remaining games, and sampled rating uncertainty. Big 12 eligible wins include nonconference results, at most one lower-division win, and exclude flagged NCAA exempt games. SEC scoring tiebreakers use the published capped-relative-margin formula, retaining final scores and modeling remaining scores conditional on each simulated winner. A random draw is used only if all preceding criteria remain tied. Pac-12 inputs stop before the flex week. Where the Big 12 or Pac-12 text does not specify how a reduction to three or more teams resumes, this forecast continues at the next criterion; it restarts after awarding a berth.
Rating uncertainty is fitted to historical CFP rankings from 2018, 2019, 2021 and 2022, then tested on 2023–2025. Among pairs of ranked teams, the proxy orders about 86% correctly in those held-out seasons; its errors are included in the simulation. SportSource historical ratings were unavailable: using the CFP-fitted uncertainty for SportSource is an unvalidated assumption, not a reproduction of its formula. The model samples one consistent rating order per season and averages qualification over those possibilities to display one probability.
50,000 seasons is a speed and precision budget. At that size, a single probability has an approximate 95% sampling margin of at most 0.44 percentage points. More simulations reduce random sampling noise, not errors in the game forecasts or rating assumptions. The conservative sampling envelope below is separate from model accuracy.
Games are sampled independently. This is not a CFP-selection model. Conference appearance does not guarantee a CFP berth. Conditional swings cannot be added together.
Model: conference-national-uncertainty-v3 · Seed: 2026
Full methodology and sources · Historical validation and limitations
Snapshot sec-2026-20261110T175559Z · Ranking inputs 73a4fecc492a7600e9f0a6f73deb6ea374c92953b4fdb94b37c730bc2f3d3425