GW 4 â Tzolakis has saved every shot. That is the whole explanation for Hull being third.
Hull City have played 294 minutes of Premier League football without conceding a goal. They create 0.69 expected goals per match, hold the lowest field tilt in the league at 0.17, and complete 183 passes per match while their opponents complete 578. Seven points, third place and three matches unbeaten, which is Hull's longest such run in the Premier League since 2016/17.
Fulham have played the same number of minutes and have zero points. They create 2.1 expected goals per match, third highest in the league, concede 1.2 and carry an expected goal difference of plus 0.9. The model gives them 4.6 expected points, which is more than Liverpool, who have five real ones. Three straight defeats.
The gap between actual and expected points is plus 2.8 for Hull and minus 4.6 for Fulham. Only Manchester City are overperforming by more than Hull, nobody is underperforming like Fulham, and on expected points Fulham actually sit above Hull, 4.6 to 4.2. Round 4 sends Hull to Stamford Bridge and Fulham to Anfield on the same afternoon. It is hard to think of two better places for a table to start correcting itself.
Ten shots, ten saves, 1.27 expected goals
The explanation for Hull's clean sheets is Konstantinos Tzolakis. He has made ten saves from ten shots on target, and those shots were worth 1.27 expected goals. He has prevented 100 per cent of what he has faced, a number that cannot get any better. The only one to match it is Sunderland's Nordi Mukiele, from three shots. Hull allow 1.4 expected goals per match and 1.6 in expected threat from open play, the most in the league. This is a team that offers up chances and has a goalkeeper who has so far shut every one of them.
Chelsea are the worst possible opponent to test that formula against. They create 1.9 expected goals per match and have the best chance creation in the league, 0.07 expected goals per attack that reaches the final third. Eight goals in three matches, and Jorrel Hato is on 0.83 assists per match from left back. Chelsea have also conceded seven, and Emiliano MartĂnez is making 6.0 saves per match, the most in the league. Two busy goalkeepers in the same fixture. The two previous Premier League meetings both ended 2â0 to Chelsea, and Hull have yet to score against them in this data series.
Fulham face a Liverpool side that has just ended its longest wait
Liverpool went six matches without a win between 3 May and 29 August. That is the club's longest winless run in the Premier League since 2016/17, and it ended in round 3. Five points from 3.8 expected is still a slight overperformance, and 1.6 expected goals conceded per match is worse than Fulham's 1.2. But they have scored in 20 consecutive matches, have won 71 per cent of their home games since 2016/17, and have the most dangerous forward in the league in Alexander Isak, 0.88 expected goals per match.
Fulham answer with two players in the top seven of the same list. CĂ©sar Palacios is third on 0.66 and Gonzalo GarcĂa seventh on 0.59. The side is classified as possession control but plays 38 long balls per match, third most, and 45.4 switches of play, second most in the league. It is a possession team that plays long and wide, creates a lot, and has not yet been paid for it. The twelve meetings since 2018 have produced 3.2 goals per match. That number makes Saturday's fixture the best of the round on paper.
Arsenal can overtake themselves at Sunderland
Arsenal have won eight in a row. That equals the club's longest winning run in the Premier League since 2016/17, set between January and March 2024, and a win at the Stadium of Light on Saturday night sets a new one. The numbers match the run: 0.33 expected goals conceded per match is the best defence in the league by a distance, one goal conceded in three matches, 6.5 expected points is the highest in the league, buildup efficiency of 0.35 is the best in the league, and they have won first touch on 100 per cent of their own defensive corners. Christos Tzolis is on 3.4 shot assists per match, second in the league.
Sunderland have two straight home wins and concede 1.1 expected goals per match, fifth best in the league. Four meetings since 2016 have produced three Arsenal wins, one draw and 3.5 goals per match. Arsenal travel without William Saliba, who is out with a back injury and no return date.
The derby is two equal attacks and two different defences
Manchester United and Manchester City create the same amount, 2.0 expected goals per match each. The difference is at the back: United allow 1.0, third best in the league, City 0.59, second best. City have the highest field tilt in the league, 0.83, and the highest effective playing time, 65.0 minutes per match. United have four straight home wins, City three straight wins and 19 consecutive matches with a goal.
The midfield duel is measurable. Kobbie Mainoo is second in the league for progressive carries, 193 metres forward with the ball per match, and second for interceptions at 12.9. The leader in both lists is Elliot Anderson, 218 metres and 15.3 interceptions, and he is on the other side. Erling Haaland has 0.82 expected goals per match, second in the league behind Isak, and receives more expected threat than anyone, 0.67 per match. Rayan Cherki leads both shot assists, 4.3, and assists, 1.1. City are without JĂ©rĂ©my Doku with a calf injury. Twenty meetings since 2016 have produced ten City wins and seven for United, and the most recent, in January, United won 2â0 at Old Trafford.
Three teams still have not scored
Aston Villa, Tottenham and Coventry have each played three rounds without scoring. Villa are the most worrying case. 0.65 expected goals per match is the lowest in the league, 15 balls won in the opposition half per match is the fewest, and JoĂŁo Gomes is suspended. Their opponents Nottingham Forest have gone six matches without a win but have 4.6 expected points against two real ones, and Igor Jesus is fifth in the league on 0.62 expected goals per match.
Tottenham create 0.95 expected goals per match and concede 2.0. They have the worst counterpressing in the league, 0.03 expected goals within eight seconds of winning the ball, and share the worst buildup efficiency, 0.29. They also draw the most fouls in the league, level with Newcastle at 16 per match, which says something about what their matches look like. Everton arrive with five points, 4.6 expected, and Thierno Barry on 0.63 expected goals per match, fourth in the league. Tottenham have won eleven of the last twenty meetings. Coventry share the worst chance creation in the league, 0.02 expected goals per attack reaching the final third, with precisely Tottenham and Aston Villa. The three goalless teams are also the three who make the least of their attacks. Coventry host Brighton, who create 2.4 expected goals per match, second most in the league, play the most long balls, 47, and make the most switches of play, 48.6.
Two teams that should worry despite the points
Crystal Palace won in round 3 and have three points. Underneath is the worst defence in the league, 2.6 expected goals conceded per match, and 0.94 expected goals conceded within eight seconds of losing the ball. Second in that list, Newcastle, are on 0.66. Palace take 0.99 corners per match and have won first touch on zero per cent of them. Jean-Philippe Mateta is out for three weeks. Ipswich have lost two in a row but have 3.7 expected points against Palace's 2.1, and Abdul Fatawu shares the league lead in open-play expected threat at 0.44 per match.
Newcastle are the other case. Five points from 2.5 expected, 0.97 expected goals per match and 2.0 conceded, and LukĂĄĆĄ HornĂÄek leads the league in goals prevented with 3.5. Monday's match at Leeds pits the two lowest effective playing times in the league against each other, 49.0 and 50.0 minutes per match. It will be the slowest match of the round.
Bournemouth against Brentford in Saturday's first slot is the round's clearest model fixture. Bournemouth have two points from 5.0 expected and five matches without a win, Brentford have the highest expected goals per match in the league at 2.6 and the best counterpressing at 0.81. Keane Lewis-Potter shares the league lead in open-play expected threat with Fatawu and wins 17 per cent of his ground duels at the same time, second worst in the league. Both numbers are true.
Konstantinos Tzolakis has prevented 1.27 expected goals out of a possible 1.27. That number cannot get better, only worse, and it is the only safe forecast about Hull City right now. Fulham, conversely, have nowhere left to fall. Three rounds in, the table is still a rumour about these teams. The question is who manages to change the rumour before it becomes a verdict.
Round 4 fixtures: BournemouthâBrentford (Sat 12/9 16.00), Aston VillaâNottingham Forest (Sat 16.00), ChelseaâHull (Sat 16.00), Crystal PalaceâIpswich (Sat 16.00), LiverpoolâFulham (Sat 16.00), TottenhamâEverton (Sat 18.30), SunderlandâArsenal (Sat 21.00), CoventryâBrighton (Sun 13/9 15.00), Man UnitedâMan City (Sun 17.30), LeedsâNewcastle (Mon 14/9 21.00). All times CEST.
Source: Premier League Weekly Report â Round 4, PlaymakerAI. Injury and suspension data checked against Premier Injuries, club channels and the Premier League suspension list on 9 September 2026.