Premier League · Gameweek 3 Verdict
Gameweek 3 Verdict: 3 from 10 as draws dominate the weekend
Published 7 September 2026 · FootyPulse AI Analysis
The AI managed just 3 correct calls from 10 fixtures in Gameweek 3, as six draws wreaked havoc on a weekend that looked clearer on paper. Season accuracy now sits at 41% (11 from 27), a reminder that home advantage and form lines mean little when defences hold firm.
The scoreboard
Three from ten. That's the blunt reality of Gameweek 3, dragging the season tally to 11 correct from 27 fixtures — 41% accuracy. The model isn't panicking, but it's a lesson: when half the weekend ends level, prediction engines struggle. Six draws from ten matches is borderline freakish; the data says it shouldn't happen this often, but the ball doesn't read spreadsheets.
Biggest upset
Brentford 1-1 Sunderland stands out as the weekend's biggest miss. The AI backed Brentford at home with 68% confidence — the highest probability pick that failed — only to watch the visitors grind out a point. Sunderland's defensive shape proved too stubborn, and Brentford's inability to convert possession into goals left the model exposed. When a near-seven-in-ten favourite draws, you feel it.
Best correct call
Ipswich Town 0-2 Liverpool delivered the best correct call, and it was earned the hard way. The model gave Liverpool just 46% to win at Portman Road — practically a coin flip — yet Arne Slot's side made it look routine with a clean sheet and two second-half goals. This was the least confident pick the AI got right all weekend, proof that even marginal probabilities can land when execution meets opportunity.
Every fixture
- Ipswich Town 0-2 Liverpool — ✅ Predicted away win (46%)
- Newcastle United 2-2 Bournemouth — ❌ Predicted home win (58%), ended draw
- Manchester City 1-0 Coventry City — ✅ Predicted home win (78%)
- Fulham 2-3 Crystal Palace — ❌ Predicted draw (38%), Palace won away
- Nottingham Forest 0-0 Tottenham Hotspur — ❌ Predicted home win (48%), ended draw
- Brighton & Hove Albion 1-1 Leeds United — ❌ Predicted home win (58%), ended draw
- Brentford 1-1 Sunderland — ❌ Predicted home win (68%), ended draw
- Hull City 0-0 Aston Villa — ❌ Predicted home win (68%), ended draw
- Everton 2-2 Manchester United — ❌ Predicted home win (48%), ended draw
- Arsenal 2-1 Chelsea — ✅ Predicted home win (52%)
Six of seven wrong calls were draws, and two of the three correct picks came in genuine toss-ups (Liverpool's 46%, Arsenal's 52%). That's variance in action. Some weekends the numbers sing; others, they mumble. The model doesn't adjust for mood — it trusts process over three-match samples. Gameweek 4 will test whether this was noise or signal. See the pre-match preview to understand how the model approached each fixture before kick-off.
Accuracy will fluctuate. The job is to stay honest about what the data said and what the pitch delivered. Three from ten isn't good enough, but 27 fixtures is still early. We track, we learn, we move.
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 3 matches.
Pre-match
3/10
30% correct
At half-time
6/10
60% correct
Where half-time changed the call
Newcastle United 2–2 Bournemouth
Pre-match: Newcastle United ✗ · Half-time: Bournemouth (2–3) ✗
Why the call changed
- Shots on target 1 v 5
- Possession 55% v 45%
Fulham 2–3 Crystal Palace
Pre-match: Draw ✗ · Half-time: Fulham (3–1) ✗
Why the call changed
- Shots on target 3 v 3
- Possession 60% v 40%
Brentford 1–1 Sunderland
Pre-match: Brentford ✗ · Half-time: Draw (1–1) ✓
Why the call changed
- Shots on target 0 v 2
- Possession 43% v 57%
Brighton & Hove Albion 1–1 Leeds United
Pre-match: Brighton & Hove Albion ✗ · Half-time: Leeds United (1–2) ✗
Why the call changed
- Shots on target 1 v 3
- Possession 68% v 32%
Hull City 0–0 Aston Villa
Pre-match: Hull City ✗ · Half-time: Draw (1–1) ✓
Why the call changed
- Shots on target 3 v 1
- Possession 33% v 67%
Everton 2–2 Manchester United
Pre-match: Everton ✗ · Half-time: Draw (1–1) ✓
Why the call changed
- Shots on target 1 v 1
- Possession 48% v 52%
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 3 fixtures for the current numbers on each match.
See how the AI called it before kickoff:
Read the Gameweek 3 preview →