refdosomething

Under construction

Calibrated probability models, scored against the market. From match outcomes to mention markets, with more to come. Grounded in the math.

scoreline_surface λ 1.62 / 1.18

away goals

home goals →

Home–

Draw–

Away–

A Dixon–Coles scoreline distribution — Poisson goals with the low-score correction. Calibrating this against bookmaker odds is the whole point.

xg_model xG 0.00

Shots0

Total xG0.00

P(≥1 goal)–

Per-shot xG summed to expected goals, and the chance of at least one. The shot model that feeds λ.

mention_market phrase "dividend"

count0

P(≥1 mention)–

P(≥2 times)–

A negative-binomial count model over past events, integrated to the probability a phrase is said. Comparing that to the market price is the point.