The Premier League’s second matchweek delivered a seven-goal thriller at Stamford Bridge, a stunning fightback at Anfield, and a five-star display from Manchester United against newly promoted Ipswich. For our forecasting model, it was a round of extremes, nailing its high-confidence goal-flow reads in tight contests but drawing a complete blank on exact scorelines and struggling to separate the contenders in the marquee fixtures.
How our forecasts did
The model’s strongest suit this weekend was the win-or-draw call, where the calls for a team to win or draw landed in eight of the ten matches. The total-goals projection performed equally well, also finding eight correct outcomes, with its firm high-confidence backing of a low-scoring affair at Sunderland proving spot on. At the other end of the scale, none of the ten exact-scoreline forecasts came off, and the half-time and full-time combination calls hit only twice. The outright match-result forecasts had a tough round, with just four of the ten calls landing.
Results
Crystal Palace 1-4 Manchester City Manchester City made it two wins from two with a dominant display at Selhurst Park, sweeping Palace aside. The model had a near-perfect read on the contest, landing its high-confidence call for a City win, along with projections for both teams to score, more than 2.5 goals, and City to score first, though its cautious 0-1 exact scoreline was blown away.
Liverpool 2-2 Nottingham Forest The round’s biggest shock unfolded at Anfield as Nottingham Forest twice took the lead to secure a point. The model’s high-confidence call for a home win was one of the afternoon’s worst misses, and its forecast for a Liverpool half-time lead and full-time win never looked likely after the visitors struck first.
Bournemouth 1-1 Everton Bournemouth’s strong historical home record against Everton could not deliver a victory, as the points were shared on the south coast. The model’s medium-confidence call for a home win was incorrect, though its projection that Bournemouth would score first proved accurate before the lead was surrendered.
Coventry 0-1 Hull City Hull City defied both the model’s expectation and recent history between these sides, grabbing an away win in a fixture that had produced three consecutive draws. The low-confidence call for a Coventry win missed the mark, but the medium-confidence view on a low-scoring contest came in comfortably.
Tottenham 0-2 Newcastle A chastening afternoon for Spurs, who are now pointless and goalless after two matches. The model’s cautious lean toward a home win was completely overturned by a ruthless Newcastle performance, and its medium-confidence calls for over 2.5 goals and both teams to score were undone by a Spurs attack missing key creators.
Chelsea 4-3 Brighton A chaotic spectacle at Stamford Bridge saw Chelsea edge out Brighton in a seven-goal classic. The model correctly called a home win and for Chelsea to score first, but its cautious view that at least one side would be kept off the scoresheet was torn apart in a breathless second half.
Leeds 1-1 Brentford Leeds’ solid recent home form wasn’t enough to break down a Brentford side that had impressed on the opening weekend. The model’s medium-confidence call for a home win was off the money, and its projection of a clean sheet for one team fell apart as both sides traded first-half goals.
Sunderland 1-0 Fulham In a contest forecast to be a tight, low-scoring affair, Sunderland took the points with a single goal. The model’s high-confidence reads on a clean sheet and fewer than 2.5 goals were its best individual calls of the round, even if its broad match-result lean toward Fulham was wrong.
Manchester United 5-2 Ipswich Manchester United roared back from an opening-day defeat with a devastating attacking display at Old Trafford. The model’s high-confidence call for a home win and its projections for over 2.5 goals and both teams to score all landed, though Ipswich snatched the opening goal against the run of play.
Aston Villa 0-1 Arsenal Arsenal ground out a crucial win at Villa Park, with a depleted Aston Villa unable to unlock a disciplined visiting defence. The model’s low-confidence call for an away win proved correct, but its expectation of goals and for both teams to score was misplaced as Villa’s injury-ravaged spine failed to fire.
What we got wrong
The model’s day-one high-confidence calls took a battering. At Anfield, Liverpool’s dominance in underlying numbers and Forest’s blunt opening-day display pointed to a comfortable home victory, but Nuno Espírito Santo’s side were clinical on the counter, punishing the prediction with an early first-half goal that flipped the contest. The analysis had noted Liverpool’s ability to generate chances, but not their vulnerability when an opponent finds a route to goal.
The second bruising miss came in north London. The model leaned toward a Tottenham win despite a nightmare season opener, reading Newcastle’s conceding of 27 shots as a weakness. The absence of Dejan Kulusevski and Xavi Simons proved decisive; without them, Spurs lacked any creative spark, surrendering the initiative and never looking like the side that was projected to score first and win. It was a case of the injuries mattering more than the defensive frailties the model had priced into Newcastle.
Across the board, exact scorelines were a write-off. From the 1-0 slog that became a seven-goal Chelsea bonanza to the cagey 0-1 called for Sunderland that flipped the other way, the ball refused to settle in the right net.
As the dust settles on Matchweek 2, the table has already taken on a jagged look. Manchester City are the only side on maximum points, Tottenham and Aston Villa’s early ambitions have been checked, and Nottingham Forest have provided the first major twist. It has been a weekend where the model’s structural reads on the game were sharp, but the precise outcomes proved as elusive as ever.
