Player Elo

1. Every season is a tournament

A season of reality TV is, underneath everything, a ranking machine: it takes a cast and produces a finish order. We treat each completed season as a round-robin tournament decided by that finish order. Every player faces every other player in their season in a head-to-head “match,” and whoever finished better wins it. Win your season and you beat the entire cast; go home first and you lost to everyone. Players who share a finish (team wins, tied eliminations) simply don't play each other. Seasons still airing are left out until they finish.

2. Matches move ratings — the Elo update

Everyone enters the system at a rating of 1000. For each match, the expected result is computed from the two players' current ratings, then the winner takes points from the loser in proportion to how surprising the result was:

expected(A beats B) = 1 / (1 + 10^((R_B - R_A) / 400))

R_winner += K × (1 - expected)
R_loser  -= K × (1 - expected)

Beating a low-rated first boot barely moves the needle. Beating a proven champion pays. The K-factor is split across the season (K = 32 / (cast size − 1)), so one season carries the same total weight whether it had 16 players or 40 — and all of a season's matches are evaluated against the ratings players brought into the season, so the order of matches within it doesn't matter.

3. One pool, three shows, twenty-six years

Seasons are processed in the order they aired, from Survivor: Borneo in 2000 through the present, across all three shows in a single shared player pool. A person keeps one rating for their entire career — crossovers carry their rating with them between shows. This is why longevity matters: winning once is impressive, but champions who keep coming back and keep beating fields full of other high-rated returnees compound their rating season after season. One-and-done early boots sink and stay sunk.

4. From rating to the 1–99 score

Raw Elo numbers are meaningless without context, so the number on a profile page is a percentile: all 1,478 rated players are ranked by rating, and each is scored 1–99 by where they sit. A 99 means almost nobody in the history of these three shows has a better competitive record. A 1 means almost everybody does.

5. What it rewards, and honest caveats

• It rewards finishing well, repeatedly, against strong fields. It does not judge strategy, edit, or likability — only results.
• Finish order is the only input, so a dominant season that ends in a blindside counts the same as limping to the same placement.
• Identity is matched by name across shows; known name changes are handled with an alias list.
• Ratings recompute from scratch whenever season data updates, so scores shift as new seasons finish.