CRED
The Champion Is Credible
Supports FELD Settled on the Field
Context
Any single football game is a coin flip with a bias toward the better team, which wins most of the time but not all of the time, and on a wet Saturday in November with a freshman quarterback the bias gets thin. A format is a machine for turning a handful of those flips into a champion. Some formats do this well, and some produce a champion that nobody, including the champion, quite believes in.
Problem
A pure knockout has high variance, because one bad Saturday puts the best team in a fifth-place game, while pure standings have low variance and little drama, and nine games is still a small sample. A format that decides the title on three games has a sample problem however exciting it is, and a format that decides it on nine games where six are dead has the same problem with worse television.
Forces
- Credibility pulls against drama. The upset is the best thing the sport sells, and too many of them turn the title into noise.
- Credibility pulls against consequence. The more games bear on the title, the less each one bears.
- Credibility pulls against mercy. A second chance helps the strong team recover, and it also helps a weak team get hot.
- Parity changes the answer. The same format behaves differently when one team is a juggernaut than when eight are even, so a format should be judged under both.
Solution
Count, for the eventual champion, how many of its games bore on the title, and prefer formats where that number is near nine. Then simulate every format under the same strength draws, at several settings of parity, and ask how often the champion was one of the strongest teams. Show the parity slider on the site. Prefer formats whose credibility holds across the slider rather than formats tuned to one world.
Therefore
Make most of a champion's games part of the case for the title, and check that the strong usually win it.
Patterns
- SSU, Strength Survives Upsets. How often the champion is among the strongest, at each parity.
- SMP, Sufficient Sample. How many of the champion's games bore on the title.
Measure
Taken together: the share of seasons in which the champion is one of the three strongest teams, and the number of its games that counted toward the title. A format honors this principle when the strong usually win and they had to win a lot to do it.
Lore
Nebraska came into the first Big 12 title game in 1996 undefeated and ranked third and lost to a Texas team with three losses, and that was the end of its national title. Kansas State did the same thing two years later in double overtime against Texas A&M. A single extra knockout game at the end of a season is the highest-variance thing a conference can add, and the people who add it should know what it costs.
Supported by
SSU Strength Survives UpsetsSMP Sufficient Sample
Seen in
| scenario | language | simulated | value |
|---|---|---|---|
| status-quo | + honors | ||
| pool-play | + honors | ||
| two-cups | ~ strains | ||
| two-cups-b | ~ strains | ||
| four-cups | ~ strains | ||
| pools-playoff | + honors | ||
| mini-playoff | ~ strains | ||
| chase | + honors | ||
| seven-settle | + honors | ||
| seven-final | + honors | ||
| seven-semis | + honors | ||
| eight-final | + honors |
- status-quo: honors. The champion comes from nine-game standings.
- pool-play: honors. Two round robins are played before the title is settled.
- two-cups: strains. The title goes through knockout rounds, softened by the second cup.
- four-cups: strains. The title goes through knockout rounds, softened by the second stage.
- pools-playoff: honors. It takes six games to qualify and three to win.
- mini-playoff: strains. Knockout pods lead into a knockout.
- chase: honors. Nine games count in the standings, with the hard ones at the end.