Premier League · Gameweek 9 Verdict
Gameweek 9 Verdict: 7/10 correct as Brighton stun City
Published 3 November 2026 · FootyPulse AI Analysis
The AI called seven of ten fixtures correctly in Gameweek 9, a solid return that nudges the season accuracy up to 46% (40 from 87). Brighton's win at the Etihad proved the weekend's defining shock, while a brave draw prediction at Villa Park came good against longer odds.
The scoreboard
Seven correct calls from ten fixtures represents a strong weekend for the model. The 70% accuracy rate lifts the season-long performance to 46%—40 correct picks from 87 matches across nine gameweeks. That's a modest but steady climb after earlier volatility, and the data suggests the model is finding its rhythm as sample size grows.
Biggest upset
Manchester City 0–1 Brighton & Hove Albion delivered the weekend's headline shock. The model backed City at home with 48% confidence—a relatively even call, but still a clear preference for the champions. Brighton's clinical away win flipped the script entirely, underlining how fine the margins remain in the Premier League's elite tier.
Best correct call
Aston Villa 1–1 Fulham was the model's least confident correct prediction, backing the draw at just 38%. Three-way outcomes always carry lower probabilities by definition, so nailing this one—against two in-form sides—represents a strong read of underlying form. Villa and Fulham shared the points exactly as forecast, validating the model's narrow preference for parity.
Every fixture
- Chelsea 2–0 Manchester United — Correct (predicted home, 58%)
- Hull City 2–2 Ipswich Town — Incorrect (predicted home, 38%)
- Coventry City 3–1 Sunderland — Correct (predicted home, 48%)
- Manchester City 0–1 Brighton & Hove Albion — Incorrect (predicted home, 48%)
- Brentford 2–0 Nottingham Forest — Correct (predicted home, 48%)
- Bournemouth 2–0 Leeds United — Correct (predicted home, 48%)
- Tottenham Hotspur 2–0 Crystal Palace — Correct (predicted home, 58%)
- Aston Villa 1–1 Fulham — Correct (predicted draw, 38%)
- Liverpool 0–1 Arsenal — Correct (predicted away, 42%)
- Newcastle United 0–1 Everton — Incorrect (predicted home, 35%)
Closing thoughts
Three misses—Hull/Ipswich, City/Brighton, and Newcastle/Everton—all featured relatively tight pre-match probabilities, reflecting genuine competitive balance rather than model blind spots. The City result will dominate headlines, but the model's read of the fixture was appropriately cautious at 48%. Seven from ten keeps momentum building into the next round of fixtures. See the pre-match preview for the original projections and probabilities.
Half-time re-predictions this gameweek
At half-time the AI re-reads each match and re-predicts. Here is how those calls did against the pre-match calls, on the same set of Gameweek 9 matches.
Pre-match
6/9
67% correct
At half-time
5/9
56% correct
- 1 finished match had no half-time re-prediction at all.
Where half-time changed the call
Hull City 2–2 Ipswich Town
Pre-match: Hull City ✗ · Half-time: Ipswich Town (0–2) ✗
Why the call changed
- xG 0.2 v 0.5
- Shots on target 0 v 2
- Possession 30% v 70%
Brentford 2–0 Nottingham Forest
Pre-match: Brentford ✓ · Half-time: Draw (1–1) ✗
Why the call changed
- xG 0.2 v 0.4
- Shots on target 0 v 0
- Possession 47% v 53%
Aston Villa 1–1 Fulham
Pre-match: Draw ✓ · Half-time: Aston Villa (2–0) ✗
Why the call changed
- xG 1.2 v 0.6
- Shots on target 2 v 1
- Possession 47% v 53%
Newcastle United 0–1 Everton
Pre-match: Newcastle United ✗ · Half-time: Everton (0–1) ✓
Why the call changed
- xG 0.2 v 0.8
- Shots on target 0 v 0
- Possession 45% v 55%
This is a single gameweek — far too small a sample to read anything into, in either direction. See the running pre-match vs half-time comparison on our accuracy page for the season picture.
Share this verdict:
Prediction figures in this article reflect the data available at the time of publication and may have shifted since. See the live Gameweek 9 fixtures for the current numbers on each match.
See how the AI called it before kickoff:
Read the Gameweek 9 preview →