Simulating the season…

ESOC Premier League · Season 6

Who lifts
the trophy?

We replay the rest of the tournament ten thousand times, using every player's real tournament results — then count who wins. The odds update live as games finish.

Live — re-checks the score sheet every minute

Championship odds

How often each team wins the title across 10,000 simulated seasons. The smaller "final" number is their chance of reaching the Grand Final at all.

Odds over time

Every finished series nudges the odds. This is the story so far.

How the forecast works

The full math lives in the methodology.

1

Rate players by real tournament games

We look at 3,100+ tournament matches over the last five years. A player's rating is calculated by combining tournament performance with ladder elo.

2

Replay the season 10,000 times

Every remaining game is simulated with those ratings, plus each pairing's personal head-to-head history. Standings, tiebreakers, and the Grand Final play out in full, every time.

3

Update the moment a game ends

The page re-checks the official score sheet every minute and re-runs the whole forecast. Results are immediately reflected in the live odds.

Projected standings

Teams are ranked by total games won — every game counts, even in a lost series. "GW now" is already banked; the projection shows where each team likely finishes, and the gold cells show how often they land in each place.

Grand Final outlook

The top two teams meet in a 35-game Grand Final (5 series of play-all-7). Click a row to expand its scoreline distribution, or pin a what-if score.

Match center

Each weekly match is five one-on-one series, and all five games of every series get played. The two-color bar splits each matchup by the chance of winning a single game. Click on any series row to expand it and view the scoreline distribution. Curious? Pin a "what if" score on any unfinished series and watch the title odds react — or read the Week 1, Week 2, Week 3, and Grand Finals analyst previews.

Statistical Note: The predicted scoreline distribution is heavy-tailed, not a standard binomial distribution. This is because player form (strength draw) is sampled once per 5-game series rather than treating each game as independent and identically distributed (i.i.d.).

Player ratings

Two numbers per player: their public ladder rating, and our tournament rating — what their actual tournament results say they're worth. Big green or red gaps are where the model disagrees with the ladder.

Tournament Elo rankings

Every player in our rating network — including stars who aren't in EPL 6 — ranked by implied tournament strength. Players with few games sit close to their "anchor" rating, so use the games filter for the proven list.

# Player EPL 6 Tourn. Elo Anchor Δ Fit games W–L

The data, in the open

Everything the model eats, browsable. Spot an error? It's fixable at the source — one already was.

FFP ladder ratings snapshot

Historical tournament games

Every scraped 1v1 across 50+ DE-era tournaments — including games vs players outside EPL 6 (dimmed), which act as rating bridges in the Elo fit. Head-to-head records use only roster-vs-roster games. Recent games weigh more (half-life adjustable in model controls).

Date Event Player 1 Player 2 Winner

Model controls — for the stats nerds

Each player gets an implied tournament Elo: a Bradley–Terry network fit over all scraped tournament games (recent games weigh more — adjustable half-life above), anchored to their ladder Elo — so ladder throwers rise and ladder-only specialists fall toward their demonstrated tournament level. Each simulation samples true strength from N(tournament Elo, σ) and plays out all remaining games; per-game probabilities additionally blend the pair's head-to-head record. Completed series are locked at their real scores.