the season simulator
The title race, top four and relegation fight, simulated.
The real table after 9 matchweeks, the 290 fixtures still to play, and 10,000 seasons counted. Arsenal win the title in 56% of them. Pick any result and the whole table moves. Checked 4 November 2026.
Projected table
10,000 seasons from the table as it stands. Ordered by expected points.
Baseline
| # | Team | Pts | Expected | Title | Top 4 | Relegated |
|---|---|---|---|---|---|---|
| 1 | Arsenal | 24 | 82.0 | 56% | 98% | 0% |
| 2 | Manchester City | 21 | 77.7 | 27% | 93% | 0% |
| 3 | Brighton & Hove Albion | 19 | 73.9 | 12% | 84% | 0% |
| 4 | Liverpool | 15 | 69.3 | 3% | 60% | 0% |
| 5 | Chelsea | 16 | 66.8 | 2% | 42% | 0% |
| 6 | Aston Villa | 12 | 58.0 | <1% | 7% | <1% |
| 7 | Brentford | 14 | 57.6 | <1% | 6% | <1% |
| 8 | Tottenham Hotspur | 12 | 56.4 | <1% | 5% | <1% |
| 9 | AFC Bournemouth | 12 | 55.1 | <1% | 3% | <1% |
| 10 | Newcastle United | 11 | 53.9 | <1% | 2% | <1% |
| 11 | Everton | 13 | 51.8 | 0% | 1% | 1% |
| 12 | Manchester United | 7 | 47.9 | 0% | <1% | 2% |
| 13 | Leeds United | 11 | 44.8 | 0% | <1% | 6% |
| 14 | Fulham | 9 | 44.2 | 0% | <1% | 6% |
| 15 | Crystal Palace | 7 | 40.2 | 0% | 0% | 19% |
| 16 | Nottingham Forest | 6 | 39.3 | 0% | 0% | 22% |
| 17 | Hull City | 11 | 36.7 | 0% | <1% | 37% |
| 18 | Ipswich Town | 11 | 36.1 | 0% | 0% | 41% |
| 19 | Coventry City | 9 | 30.1 | 0% | 0% | 80% |
| 20 | Sunderland | 5 | 28.6 | 0% | 0% | 85% |
Range is the middle 80% of simulated seasons. Percentages are the share of seasons, not a forecast of one.
Pick the results
Tap a result for any fixture and the table above updates. Tap again to clear it. A picked result counts as 1–0, 1–1 or 0–1.
Everton v Coventry City
Fri 6 Nov
Leeds United v Tottenham Hotspur
Sat 7 Nov
Arsenal v Hull City
Sat 7 Nov
Fulham v Newcastle United
Sat 7 Nov
Nottingham Forest v Manchester City
Sat 7 Nov
Ipswich Town v AFC Bournemouth
Sun 8 Nov
Sunderland v Chelsea
Sun 8 Nov
Crystal Palace v Liverpool
Sun 8 Nov
Brighton & Hove Albion v Brentford
Sun 8 Nov
Manchester United v Aston Villa
Sun 8 Nov
Nudge a club
Think a club is better or worse than the model has it, or a key player is out for the run-in? Move its rating and watch the table respond. It shifts every remaining fixture for that club.
Rating 0.80 (model 0.80)
The working
Every club has one rating on a log-goals scale. It starts from a pre-season prior, then moves by how far its results so far beat or missed what that prior predicted. The prior is worth 24 matches, so nine games move a rating about a quarter of the way and more games move it further.
A match turns the two ratings and a home advantage into an expected-goals pair. Goals are drawn from a Poisson distribution, three points go to the winner and one each for a draw. The table is ranked on points, goal difference, then goals scored, and the season is played 10,000 times with a fixed seed, so the same inputs always give the same page.
It is a starting point to argue with, not an edge over anyone. It knows nothing about injuries, transfers or form beyond results, which is what the nudge is for. The same engine, with the prior as the only input, runs the from-scratch season simulation.
Questions about the simulator
Who will win the Premier League in 2026/27?
After 9 matchweeks and 10,000 simulated seasons, Arsenal win the title in 56%, Manchester City win the title in 27%, Brighton & Hove Albion win the title in 12% and Liverpool win the title in 3%. That is the share of simulated seasons, not a prediction of the one that will be played.
Who will finish in the top four?
Arsenal 98%, Manchester City 93%, Brighton & Hove Albion 84%, Liverpool 60%, Chelsea 42% and Aston Villa 7%. Four places are filled in every simulated season, so these chances add up to 400% across the league.
Who is most likely to be relegated?
Sunderland 85%, Coventry City 80%, Ipswich Town 41%, Hull City 37% and Nottingham Forest 22%. Three clubs go down in every simulated season, so the relegation chances add up to 300%.
How does the Premier League simulator work?
It starts from the real table, then plays the 290 fixtures still to come 10,000 times. Each club has a rating, set from a pre-season prior and moved by its results so far. A match turns two ratings and home advantage into expected goals, goals are drawn from a Poisson distribution, and points fall out. The same seed always gives the same answer.
Is this betting advice?
No. These are probabilities from a transparent model, built to argue with. They are not tips, and a probability is a statement about many seasons, never a guarantee about one.