How Scorient Works

Dixon-Coles, Pi-Rating, and xG — the mathematics behind football intelligence.

Step 1

Data Collection

The system gathers football data from multiple sources — including match results, team and player statistics, form trends, injuries, and news. This data is continuously updated so the AI has access to the latest information before generating predictions.

Step 2

Dixon-Coles Model Processing

Scorient's core prediction engine uses the Dixon-Coles statistical model — a peer-reviewed framework published in 1997 that improved on the standard Poisson distribution by introducing a correction parameter for low-scoring scorelines (0-0, 1-0, 0-1, 1-1). The model estimates attack and defence strength parameters for each team from historical results, then generates score probability distributions for upcoming matches. This approach is applied across:

  • Historical match data — result patterns, home advantage factors, and form weighting.
  • Team attack/defence ratings — separately calibrated home and away parameters.
  • Low-score correction — the signature Dixon-Coles rho parameter improving draw prediction.

Step 3

Pi-Rating & xG Overlay

Alongside Dixon-Coles, Scorient applies Pi-Rating — a dynamic Elo-style team rating system calibrated specifically for football — to provide a real-time team strength signal that updates faster than parameter-based models. Expected Goals (xG) data is then overlaid to detect teams currently over- or under-performing their underlying shot quality. This three-layer approach gives Scorient a more complete picture: long-run historical modelling (Dixon-Coles), current form signal (Pi-Rating), and efficiency analysis (xG).

Step 4

Prediction Output & Subscriptions

The combined model outputs probability estimates for Home Win, Draw, and Away Win — as well as both-teams-to-score, projected scorelines, and value bet signals (Pro and Elite plans). All outputs are labelled with confidence tiers so you always understand the uncertainty level. Access is via subscription: Free (5 predictions/day), Basic ($4.99/mo), Pro ($12.99/mo, includes xG), and Scorient Elite ($29.99/mo, includes value bets and API access).

Step 5

Continuous Learning

Models improve over time using new football data. As more matches are played and more statistics become available, the system refines its algorithms to deliver better analytical insights. Scorient is built to learn and adapt — so the more data we process, the sharper the analysis.