AI Premier League predictions — graded in public
England · Updated
As of 27 October 2026, 10 upcoming Premier League fixtures are scheduled in the next few days, and ScoreGPT's AI panel has not analyzed them yet — the picks publish about 36 hours before each kickoff and are graded publicly after full-time. Across the last 150 graded Premier League matches ScoreGPT covered, the consensus called 69 correctly.
Download the free app
3 free AI predictions every week — no sign-up.
Upcoming Premier League fixtures
Matchday 9
AI analysis lands about 36 hours before each kickoff — picks appear here as the models finish.
Predicted final Premier League table — after matchday 7
After matchday 7, the five AI models' combined Premier League table has Manchester City top on a projected 86 points, with Nottingham Forest, Coventry and Sunderland in the bottom three.
| # | Team | Proj. pts | Proj. GD |
|---|---|---|---|
| 1 | Manchester City | 86 | +50 |
| 2 | Arsenal | 82 | +40 |
| 3 | Brighton | 69 | +28 |
| 4 | Chelsea | 68 | +20 |
| 5 | Liverpool | 68 | +22 |
| 6 | Manchester United | 56 | +6 |
Places 7–17
| # | Team | Proj. pts | Proj. GD |
|---|---|---|---|
| 7 | Aston Villa | 56 | +4 |
| 8 | Brentford | 54 | +7 |
| 9 | Newcastle | 54 | +4 |
| 10 | Tottenham | 51 | 0 |
| 11 | Leeds | 49 | 0 |
| 12 | Everton | 49 | -1 |
| 13 | Bournemouth | 44 | -8 |
| 14 | Crystal Palace | 43 | -12 |
| 15 | Hull City | 42 | -10 |
| 16 | Ipswich | 41 | -13 |
| 17 | Fulham | 40 | -12 |
| # | Team | Proj. pts | Proj. GD |
|---|---|---|---|
| 18 | Nottingham Forest | 38 | -14 |
| 19 | Coventry | 36 | -22 |
| 20 | Sunderland | 33 | -26 |
- Champions League
- Europa League
- Relegation
Projected points are final totals: the points each club has already won plus what the models expect from its remaining games.
Each model's calls
| Model | Champion | Europe (places 1–5) | Relegation (places 18–20) |
|---|---|---|---|
| ScoreGPTcombined table | Manchester City | Manchester City · Arsenal · Brighton · Chelsea · Liverpool | Nottingham Forest · Coventry · Sunderland |
| GPT-6.1 Sol | Manchester City | Manchester City · Arsenal · Chelsea · Liverpool · Brighton | Ipswich · Sunderland · Coventry |
| Claude Opus 5.5 | Manchester City | Manchester City · Arsenal · Liverpool · Brighton · Chelsea | Ipswich · Coventry · Sunderland |
| Grok 4.7 | Manchester City | Manchester City · Arsenal · Brighton · Chelsea · Liverpool | Fulham · Coventry · Sunderland |
| Kimi K3 | Manchester City | Manchester City · Arsenal · Liverpool · Chelsea · Brighton | Coventry · Nottingham Forest · Sunderland |
| DeepSeek-V4-Pro | Arsenal | Arsenal · Manchester City · Brighton · Liverpool · Chelsea | Coventry · Nottingham Forest · Sunderland |
The ScoreGPT row is not another AI model: it is a consensus synthesized from the five models' tables after they answered.
How these tables were made: after matchday 7, each of the five models — GPT-6.1 Sol, Claude Opus 5.5, Grok 4.7, Kimi K3 and DeepSeek-V4-Pro — was given the real Premier League table, with every club's points, games played and games left, and asked for the complete final table, using its own knowledge of the clubs with web search switched on. The tables come from the five models' APIs; no model saw another model's table.
Every club's reasoning, model by model, is in the ScoreGPT app.
The forecast made before matchday 1 →Image of this table (free to use with a credit)
Cite this table
Source: ScoreGPT, five-AI predicted Premier League table after matchday 7 (20 October 2026), scoregpt.app/ai-football-predictions/premier-league#ai-predicted-table
Graded Premier League record
ScoreGPT’s consensus record: 69/150 (46.0%) results correct across the last 150 graded Premier League matches ScoreGPT covered · 14/30 (46.7%) in the last 30 days. Misses included — see the per-model leaderboard. The 40 most recent graded matches are shown below.
- FTSunderland0Leeds0✗ MissAI: Leeds
- FTCrystal Palace0Newcastle2✓ HitAI: Newcastle
- FTHull City1Brentford1✗ MissAI: Brentford
- FTLiverpool2Brighton0✓ HitAI: Liverpool
- FTManchester United0Bournemouth3✗ MissAI: Manchester United
- FTChelsea0Tottenham2✗ MissAI: Chelsea
- FTCoventry0Fulham1✓ HitAI: Fulham
- FTArsenal2Everton0✓ HitAI: Arsenal
- FTAston Villa2Manchester City1✗ MissAI: Manchester City
- FTIpswich1Nottingham Forest1✗ MissAI: Nottingham Forest
Earlier results30
- Tottenham 3–1 Coventry ✓ Hit
- Nottingham Forest 0–1 Arsenal ✓ Hit
- Leeds 1–1 Manchester United ✗ Miss
- Bournemouth 2–0 Sunderland ✓ Hit
- Brighton 5–0 Crystal Palace ✓ Hit
- Newcastle 0–1 Aston Villa ✗ Miss
- Brentford 1–2 Liverpool ✗ Miss
- Fulham 1–0 Hull City ✓ Hit
- Manchester City 3–2 Ipswich ✓ Hit
- Everton 0–2 Chelsea ✗ Miss
- Coventry 4–1 Newcastle ✗ Miss
- Liverpool 0–3 Manchester City ✓ Hit
- Crystal Palace 3–1 Nottingham Forest ✗ Miss
- Hull City 2–2 Everton ✗ Miss
- Manchester United 1–1 Tottenham ✗ Miss
- Aston Villa 2–2 Brentford ✗ Miss
- Chelsea 5–1 Bournemouth ✓ Hit
- Ipswich 2–1 Fulham ✗ Miss
- Sunderland 0–2 Brighton ✓ Hit
- Arsenal 2–1 Leeds ✓ Hit
- Fulham 1–1 Manchester United ✗ Miss
- Bournemouth 0–1 Liverpool ✓ Hit
- Leeds 0–0 Crystal Palace ✗ Miss
- Manchester City 5–3 Sunderland ✓ Hit
- Nottingham Forest 0–1 Coventry ✗ Miss
- Everton 1–0 Ipswich ✓ Hit
- Brighton 3–0 Arsenal ✗ Miss
- Newcastle 2–1 Hull City ✓ Hit
- Tottenham 2–3 Aston Villa ✗ Miss
- Brentford 3–0 Chelsea ✓ Hit
How the Premier League predictions work
When the Premier League (the EPL, to many soccer fans outside the UK) is in season, every covered fixture is read independently by multiple frontier AI models — form, injuries, lineups, congestion from midweek European trips — and combined into one consensus prediction with a predicted scoreline, published about 36 hours before kickoff.
The Premier League punishes lazy assumptions: bottom-three sides beat title challengers every season. That is why every pick here is graded in public after full-time and stays on the page, hits and misses alike — the record below is the models' actual output, not a highlight reel.
The process is documented in the methodology and every model's history is on the accuracy leaderboard.
Season forecast: every model's predicted 2026/27 Premier League table →
Frequently asked questions
Frequently asked
▸How do the AI Premier League predictions work?
Several large language models analyse each Premier League match independently — form, injuries, lineups and matchup context — and ScoreGPT combines their picks into one consensus prediction. Every pick is graded publicly after the final whistle.
▸Are the Premier League predictions free?
The consensus pick for every covered match is free on this page. The app adds every model's individual pick and full reasoning — 3 free AI predictions per week, no sign-up.
▸How accurate are the AI's Premier League predictions?
Accuracy varies by model and by matchweek, and no AI can guarantee results. ScoreGPT publishes the graded record — including misses — on this page and at scoregpt.app/accuracy.
▸When do predictions for a Premier League matchweek appear?
About 36 hours before kickoff, when a fixture enters the analysis window.
For information and entertainment only.