Football Predictor by Martin

One-page results-based predictor using recent form, venue strength and Poisson scoreline probabilities.

✓ Client-side only CSV source: Football-Data.co.uk No API key

1. Load results

Or drag a Football-Data CSV file here if browser fetch is blocked.
Choose a league and click Load results. Browser CORS rules can be awkward, so there is a proxy fallback and a CSV upload option.
Matches loaded
—
Completed full-time results
Teams
—
Detected from CSV
Avg goals
—
Home / away
Latest result
—
Date in file

Model settings

Strong opinion: recent form alone is a weak predictor. This tool uses recent results, but it deliberately smooths them towards league averages so one mad 5–0 does not hijack the forecast.

2. Pick a fixture

v
Prediction will appear once results are loaded and teams are selected.
How the model works

Data: the app expects Football-Data style CSV columns: Date, HomeTeam, AwayTeam, FTHG, FTAG and FTR.

Team strength: it calculates weighted recent goals for and against, with home and away splits. It smooths each team back towards the league average so small samples behave sensibly.

Expected goals: home xG is based on home attack × away defence × league home scoring rate. Away xG is based on away attack × home defence × league away scoring rate. A small recent goal-difference boost can be applied.

Probabilities: it uses a Poisson distribution to estimate scoreline probabilities, then sums those into home win, draw and away win.

Limits: this does not know injuries, lineups, red-card effects, fixture congestion, tactical changes or motivation. It is a transparent baseline, not a betting oracle.