Ohms Premier League is a quantitative match-intelligence engine. Every number on this site is produced by a deterministic model — the language model narrates, it never sets a probability or a rating.

1 · Elo. Each club carries a strength rating updated after every result. Home advantage and margin of victory feed the update. This is the prior the match model leans on.

2 · Dixon-Coles. The two ratings convert into expected goals for each side, then into a bivariate-Poisson scoreline distribution — the 7×7 heat grid on each match page. Low-score correlation is corrected the Dixon-Coles way so 0–0 / 1–1 aren't overstated.

3 · Monte-Carlo. The remaining fixtures are simulated tens of thousands of times to produce the title / top-four / relegation probabilities on the standings, and each club's expected points.

4 · De-vig & conviction. Where a market price exists, the bookmaker overround is stripped out to a fair probability, compared against the model, and any edge is sized by half-Kelly. That's the pundit board.

5 · Signals. Bedrock extracts facts from the news — injuries, suspensions, form — and a bounded rules engine turns them into small, traceable Elo adjustments. Every adjustment carries its source URL and rule version. Nothing the LLM writes moves a number directly.

Same void/green/gold engine as the Ohms World Cup console — only the competition differs.