Supercomputer · Premier League

Plays the rest of the season 10,000 times and shows how often each club wins.

Loading data…

The Premier League supercomputer, explained

Twenty clubs, 38 matches, and the sharpest cut-off on the site: the top four go into the Champions League league phase. A fifth English place appears whenever England finishes among the two best-performing associations in Europe, as it has in each of the last two seasons, but that cannot be assumed in August.

Below the Champions League line, the highest finisher outside it takes a Europa League place alongside the FA Cup winner, while the League Cup winner enters the Conference League play-off round. Cup winners who already qualify push those berths down the table, so fifth and sixth are worth different things in different seasons. Three clubs go down.

Premier League projection: frequently asked

How many Premier League clubs qualify for Europe?

Seven as a baseline: four in the Champions League, two in the Europa League and one entering the Conference League play-off round. That rises to eight in seasons when England earns one of UEFA's two European Performance Spots, and cup winners already qualified through the league can push places further down.

Can the projection account for points deductions?

No. From 2026/27 the league replaced its profitability rules with a Squad Cost Ratio, under which a serious overspend brings a six-point deduction that scales with the size of the overspend. Rulings like that land mid-season and sit entirely outside the model, so read the relegation odds as football only.

What does the top-six column actually mean here?

It is a reference marker, not a prize in itself. Fourth is the Champions League line and the number most readers want. Fifth and sixth usually lead to the Europa League or Conference League, but in some seasons sixth brings nothing extra at all, depending on who wins the domestic cups.

▸How does the supercomputer work?

Every remaining fixture is simulated with a Poisson model: each club gets an attack rating and a defence rating from the goals it has scored and conceded this season, and the expected goals of a match come from combining one side's attack with the other's defence, plus a moderate home advantage.

Before a ball is kicked the model does not start from zero: each club begins with a prior strength built from last season's finish and the market value of its squad, with the tails capped so no extreme is absurd. That starting point hands over weight to real results as they arrive, and by around matchday ten only the actual goals are talking.

The remaining season is played 10,000 times in a separate browser process, so the page never freezes. The title, top-four, top-six and relegation percentages are simply how often each outcome happened across those 10,000 seasons, and each club's points range covers the central 90% of its simulated finishes.

What the model does not know: injuries, transfers, sanctions, European fixtures or motivation. It uses goals and only goals — which is why it is in beta.