← See Full Rankings

About Matchday Insights

One rating model, every tracked league on earth, on a single comparable scale. A fourth-tier German side and Real Madrid are measured the same way — so the question "how far apart are they, really?" has an actual number behind it.

What this is

Most football rankings cover one league, one continent, or one competition. Matchday Insights covers thousands of clubs across every tracked division and puts them all on one scale. The rating a club carries here means the same thing whether they play in the Premier League, the Ekstraklasa, or the Mongolian Premier League.

Every senior competitive result feeds the model: domestic leagues, domestic cups, continental competitions, and global tournaments. Youth, reserve and women's competitions are tracked separately and excluded from the senior ranking.

How the rating works

The engine is an Elo model. Every club carries a single number. Beat a stronger opponent and you gain more than you would for beating a weaker one; lose to a weaker opponent and you lose more. The size of the change also reflects the margin of victory and the importance of the competition.

Ratings are rebuilt from scratch on every update, replaying every match in order from 2020 onward. Nothing is incremental, so a correction to an old result propagates properly through everything that followed it.

Connecting the leagues

The hard part of a global ranking is not rating clubs within a league — it is working out how leagues compare to each other. Two mechanisms handle it:

Context that matters

Match forecasts

Upcoming fixtures carry win, draw and loss probabilities. Elo alone cannot produce these — its expected score blends winning and drawing into one number (P(win) + 0.5 × P(draw)) and never separates them.

The split uses Davidson's extension of the Bradley–Terry model, with the draw parameter fitted by maximum likelihood on this project's own match history rather than assumed. Draws are most likely between evenly matched sides and fall away as the gap widens, which is what actually happens.

How accurate is it? Fitted on 220,495 historical matches and tested on matches it had never seen, the model predicted draws at 25.1% against an actual rate of 25.3%. Across probability bands from 8% to 30%, predicted and actual rates tracked within roughly a point.

Continental football behaves differently enough — fewer draws, and a much flatter relationship with rating gap — that it gets its own fitted parameters.

Updates

The site refreshes daily. Ratings are recalculated from every result available at the time, and the rankings, club pages, forecasts and history charts are all regenerated together.

Data

Match results, fixtures and league standings come from API‑Football. League standings are read directly rather than inferred from results, so points deductions and other administrative adjustments are reflected correctly.

Who makes it

Matchday Insights is built and maintained by one person — the rating engine, the data pipeline and the site. If something looks wrong, or a club or competition is missing, get in touch.