Prediction Methodology

Key facts. LiveGoalz publishes model-estimated probabilities for the three full-time outcomes of a football match — home win, draw, away win. Model version det-2026-06-15b. Forecast horizon: pre-match. Coverage: major club and international competitions, including the FIFA World Cup 2026. Last updated 2026-06-17. The model is newly deployed; a public, settled track record is being collected and will be published on the performance page.

What we forecast

For each covered fixture the model outputs three probabilities — home (1), draw (X) and away (2) — that always sum to 100%. These are estimates of how likely each result is, not guarantees. A pick (the most likely outcome) and a confidence level are derived from the same probabilities.

Data inputs

The probabilities are derived from objective, recent team statistics:

• team scoring and conceding rates per game, taken from current-season splits;
• venue-specific form — separate home and away per-game rates;
• head-to-head history between the two teams;
• competition context (level and type of the competition).

Bookmaker 1X2 odds are shown alongside the model as a same-horizon market reference. They are a benchmark for comparison, not the source of the model’s pick — the probabilities come from the team statistics above.

Forecast horizon

Forecasts are pre-match: each published probability set is the model’s view captured at a defined point before kick-off. Live in-play probabilities are not part of this model.

Model version and reproducibility

The current model is det-2026-06-15b. The leading det marks it as deterministic: identical inputs always produce identical outputs, so every forecast is reproducible and the model’s behaviour cannot change silently. Each released version is pinned to a frozen benchmark so unintended changes are caught before deployment.

Probability calibration

Outputs are probabilities that sum to one. Calibration — whether a stated probability matches the observed frequency of that outcome over many forecasts — is measured against settled results and reported on the performance page once the settled sample is large enough to be meaningful. Until then we make no calibration claim.

Settlement rules

A forecast is settled by the official full-time result (90 minutes plus stoppage time; extra time and penalties are excluded unless a specific market states otherwise). Postponed or abandoned fixtures are voided rather than scored.

Known limitations

• the model is newly deployed and has no published track record yet;
• for teams with little recent data, confidence is capped and the forecast is labelled data-limited;
• the model reads recent form, not team news — injuries, suspensions and line-ups are reflected only insofar as they already show up in results;
• football has high variance: even correct probabilities produce many “wrong” single-match calls.

Forecasts are informational only and are not betting advice. 18+. If gambling is affecting you or someone you know, please seek support.

Changelog

• 2026-06-17 — methodology page published; model det-2026-06-15b live.