Extinction.GG is a history of high school cross-examination policy debate rounds. It covers every tournament where TOC bids are offered, from 2015–2016 to the current season.
It records win-loss results and an estimated Elo rating for ranking teams on the national circuit. That is a distinct ranking from the NDCA Baker points system and from the TOC bid system — it measures results against opponents, not accumulated bids.
The Elo algorithm uses the standard k-weight algorithm, where a 400-point difference is a 90% win-likelihood percentage. The current model uses a default rating of 1500 and a K factor of 32. You can calculate expected win rate and Elo gain here.
Design decisions are inspired by Ultimate Tennis Statistics. There is a “newbie rating adjustment” algorithm which allows teams close to the default rating to fluctuate much more aggressively: 1 + 18 / (1 + 2^((rating − 1500) / 63)) times the Elo you would have gained or lost is applied.
Ratings are calculated in chronological order. This decision is because policy is swiss matched but power matched, where all teams with the same number of wins will debate each other in preliminary rounds.
There are no repeat matches during calculations (no repeat simulations). This intends to include all TOC tournaments. Elo gains are weighted on tournament prestige: 100% on Octos bid tournaments, 90% on Quarters, 85% on Semis, and 80% on Finals.
Elo is computed within a single season and does not carry over between seasons.
The model on PolicyDebateCentral has a K factor of 16 and randomises the match ordering, then repeats the results 100 times. They include TOC tournaments with equal weighting.
Each season's leaderboard lists which bid tournaments are included, which are missing, and which are present but lost rounds. The earlier the season, the more often a tournament used software Tabroom cannot export cleanly — those are skipped and noted rather than partially ingested.