What we learned from FPL pre-season: all 20 clubs assessed

The pre-season verdict

The biggest positive team signals came from Brentford, Brighton, Ipswich and Tottenham: each combined results with repeatable role evidence, while Arsenal's Community Shield win supplied the strongest competitive selection clue. Bournemouth scored most freely but did not keep a clean sheet; Crystal Palace, Newcastle, Aston Villa and Sunderland supplied the clearest defensive or attacking warnings.

Minutes matter more than goals when translating summer matches into FPL. Haaland, Bruno Fernandes, Gabriel, Watkins, Thiago, Tavernier, Gibbs-White and several centre-backs project as genuine locks. Arsenal's front line, City's creators, Chelsea's premium attack, Fulham's full-backs and Spurs' supporting attackers remain materially less certain.

There is no complete, comparable public pre-season xGA dataset for all 20 clubs. This review therefore uses official goals conceded and clean sheets for the summer, then adds the FPLRogue model's fixture-adjusted Gameweek 1 xGA and clean-sheet probability. It does not relabel goals conceded as xGA.

Results and key moments

The club records below cover every result listed in the Premier League's completed 2026 pre-season hub through 16 August. Match counts and opponent quality vary, split-squad and shortened games are included where the hub lists them, and cross-Premier-League matches appear in each club's own record. The score totals are descriptive rather than a league table.

• Arsenal — 6 listed games, 14 goals for, 8 against and 2 clean sheets. Kai Havertz started a third successive match in the Community Shield, Christos Tzolis supplied two assists, Martin Ødegaard scored and David Raya made the saves for a 3-0 win. The model still makes Gabriel (17.17 xPts over GW1-3; 89.6 GW1 xMins) the safest route, while Gyökeres (14.35 xPts over GW1-3; 66.3 GW1 xMins) leads Havertz (7.85 xPts over GW1-3; 29.1 GW1 xMins) for minutes. That friendly-model conflict keeps the centre-forward role open; Riccardo Calafiori also carries only 63.8 xMins despite his Shield goal.
• Aston Villa — 8 games, 17 for, 14 against and only 1 clean sheet. John McGinn scored after the senior half-time changes improved the final friendly, but five defeats and 14 concessions make the defence the underwhelming unit. Watkins (14.14 xPts over GW1-3; 85.3 GW1 xMins) is the clearest attacking minutes lock; Ross Barkley, Pau Torres, Ian Maatsen and McGinn all sit below 69 xMins, so the supporting cast and both full-back slots need team news.
• Bournemouth — 5 games, 22 for, 8 against and no clean sheets. The 10-1 Genoa result inflates the attacking total, but Evanilson scoring at Mainz and Marcus Tavernier repeatedly reaching useful positions are transferable clues. Tavernier (13.23 xPts over GW1-3; 87.5 GW1 xMins) and Evanilson (11.02 xPts over GW1-3; 84.1 GW1 xMins) look secure; Rayan (10.76 xPts over GW1-3; 62.9 GW1 xMins) has upside with less certain minutes. Zero pre-season shutouts meets a brutal opening model forecast of 2.17 xGA away to Manchester City.
• Brentford — 4 wins from 4, 16 for, 4 against and 1 clean sheet. Igor Thiago and Kevin Schade scored twice each in the 7-0 final against a recognisable Frankfurt XI, with Dango Ouattara and Keane Lewis-Potter also productive. Thiago (15.37 xPts over GW1-3; 86.9 GW1 xMins) is the model's secure focal point; Schade (12.39 xPts over GW1-3; 69.0 GW1 xMins) and Dango Ouattara (12.89 xPts over GW1-3; 67.3 GW1 xMins) retain meaningful substitution risk. Nathan Collins and Michael Kayode both project above 88 xMins.
• Brighton — 4 wins from 4, 12 for, 4 against and back-to-back clean sheets against Roma and Bologna. Jack Hinshelwood scored the final goal in a repeated XI, while Maxim De Cuyper took corners and Mats Wieffer attacked aerial situations. Pascal Groß (12.22 xPts over GW1-3; 83.1 GW1 xMins) and Hinshelwood (12.06 xPts over GW1-3; 76.4 GW1 xMins) lead the model, but Wieffer (11.62 xPts over GW1-3; 55.3 GW1 xMins) remains a minutes risk. Yankuba Minteh's injury is the clearest availability downgrade.
• Chelsea — 6 games, 16 for, 11 against and 1 clean sheet. João Pedro's final-match brace and full 90 minutes are the standout late evidence; Morgan Rogers scored on his first pre-season appearance, while the defence conceded in five of six games. Enzo Fernández (13.82 xPts over GW1-3; 84.4 GW1 xMins) is the model minutes anchor, but Cole Palmer (12.72 xPts over GW1-3; 65.5 GW1 xMins) and João Pedro (12.19 xPts over GW1-3; 62.5 GW1 xMins) both sit below 66 xMins. Their ceiling is clear; their opening workload is not.
• Coventry City — 6 games, 9 for, 5 against and 3 clean sheets. Ellis Simms finished the final win over Monaco, Matt Grimes created from midfield and Loum Tchaouna also scored. Gustavo Hamer (10.24 xPts over GW1-3; 83.3 GW1 xMins) looks secure, but Simms (10.14 xPts over GW1-3; 61.8 GW1 xMins) and Thomas-Asante (10.15 xPts over GW1-3; 61.8 GW1 xMins) are level on 61.8 xMins, making striker selection a genuine battle. The three shutouts are positive; opening at Arsenal produces only a 6.9% model clean-sheet chance.
• Crystal Palace — 6 games, 7 for, 11 against and 1 clean sheet, with four scoreless outings. Freiburg led 2-0 after the scheduled 90 minutes before adding a third in an extra training period; Evann Guessand, Daniel Muñoz and Eddie Nketiah supplied the better late attacking moments without a goal. Muñoz (11.69 xPts over GW1-3; 81.0 GW1 xMins) is the secure FPL route, while Ismaïla Sarr (11.95 xPts over GW1-3; 72.4 GW1 xMins) and Canvot remain less certain. The attack and defence both underwhelmed on raw results.
• Everton — 6 games, 10 for, 6 against and 2 clean sheets, finishing with a 3-1 win over Newcastle. The team trend improved after the Stuttgart defeat, and the model treats Dewsbury-Hall (12.95 xPts over GW1-3; 87.5 GW1 xMins) and Iliman Ndiaye (12.16 xPts over GW1-3; 87.1 GW1 xMins) as near-locks. Beto (11.90 xPts over GW1-3; 66.0 GW1 xMins) carries greater substitution risk. James Tarkowski projects 87.3 xMins and the opening home fixture has a 39.9% clean-sheet probability, a stronger signal than the mixed summer sequence alone.
• Fulham — 6 games, 7 for, 8 against and 1 clean sheet. Timothy Castagne converted Alex Iwobi's corner against Stuttgart and Bernd Leno made important saves, giving the final match clearer FPL value than the overall totals. Calvin Bassey is the defensive minutes lock at 89.6 xMins; Sessegnon (12.41 xPts over GW1-3; 57.9 GW1 xMins) leads the club for xPts but is far less secure than Castagne (12.25 xPts over GW1-3; 79.7 GW1 xMins). Antonee Robinson also sits below 69 xMins.
• Hull City — 5 games, 6 for, 4 against and 2 clean sheets, but no goals in the final two. Konstantinos Tzolakis produced the standout final performance with repeated saves, while Mohamed Belloumi supplied corners, long throws and crosses without the attack converting. Oli McBurnie (10.58 xPts over GW1-3; 84.9 GW1 xMins) is the minutes lock; Belloumi (7.32 xPts over GW1-3; 45.5 GW1 xMins) is not. The defensive record looks tidy, but Manchester United's visit carries 2.97 model xGA and only a 5.1% shutout chance.
• Ipswich Town — 6 games, 13 for, 5 against, 3 clean sheets and 5 wins. Jack Clarke and Leif Davis repeatedly combined on the left in the 4-2 win at Union Berlin, with Davis providing the clearest defender upside. Davis (9.24 xPts over GW1-3; 83.4 GW1 xMins) and Dara O'Shea (9.22 xPts over GW1-3; 88.4 GW1 xMins) look like minutes locks; the model's nominal points leader, Saša Lukić, sits at only 66.9 xMins. The strong summer defence is credible enough to monitor, not enough to ignore opponent adjustment.
• Leeds — 5 games, 10 for, 6 against and 2 clean sheets, including wins over Liverpool and RB Leipzig before the draw with Manchester United. Brenden Aaronson scored in the final game; Dominic Calvert-Lewin and Harry Wilson then appeared in the second-half XI and threatened. Dominic Calvert-Lewin (12.63 xPts over GW1-3; 87.6 GW1 xMins) is the strongest model route, with Wilson at 77.2 xMins. Anton Stach (10.88 xPts over GW1-3; 68.8 GW1 xMins) and Aaronson are less secure.
• Liverpool — 6 games, 11 for, 9 against and 3 clean sheets. Cody Gakpo's goal and assist in the stronger Anfield XI against Como is the standout final attacking clue; Jérémy Jacquet scored on debut and Alexis Mac Allister made his first pre-season appearance in the earlier shortened game. Dominik Szoboszlai (13.46 xPts over GW1-3; 89.2 GW1 xMins), Gakpo (12.57 xPts over GW1-3; 84.5 GW1 xMins) and Virgil van Dijk all project above 84 xMins. Milos Kerkez at 61.0 and Jeremie Frimpong at 50.3 remain the defensive minutes risks.
• Manchester City — 4 games, 7 for, 6 against and no clean sheets, ending with a 3-0 Community Shield defeat. Erling Haaland and Phil Foden still tested Raya and Antoine Semenyo had a late effort saved, but the defence was the underwhelming unit. Haaland (24.95 xPts over GW1-3; 86.8 GW1 xMins) is the clearest lock; Nico O'Reilly and Semenyo also project above 83 xMins. Cherki (17.21 xPts over GW1-3; 67.6 GW1 xMins), Foden (14.25 xPts over GW1-3; 62.5 GW1 xMins) and Jérémy Doku at 71.2 are the high-upside rotation cluster. The model still gives City a 40.9% GW1 clean-sheet probability.
• Manchester United — 6 games, 11 for, 8 against and 1 clean sheet. Bruno Fernandes retained penalty duty but had his kick saved as United conceded three unanswered goals against AC Milan; Matheus Cunha began centrally in a fluid front three with Amad. Bruno Fernandes (22.47 xPts over GW1-3; 86.7 GW1 xMins) is a model lock, with Bryan Mbeumo and Cunha above 75 xMins. Amad, Harry Maguire and Lisandro Martínez remain below 73. The model's 0.79 GW1 xGA conflicts with the four-goal final collapse, so the fixture and personnel matter more than either signal alone.
• Newcastle United — 7 games, 10 for, 12 against and only 1 clean sheet. Malick Thiaw scored against Leverkusen and William Osula scored from Jacob Murphy's cross in the rotated Strasbourg XI, but the team won only twice and shipped four to Bristol City plus three to Everton. Thiaw (11.58 xPts over GW1-3; 89.4 GW1 xMins) is the minutes lock. Harvey Barnes, Lewis Miley and Murphy sit between 61 and 65 xMins; Osula (8.62 xPts over GW1-3; 36.9 GW1 xMins) is an impact route rather than a secure starter.
• Nottingham Forest — 9 games, 13 for, 9 against and 3 clean sheets. Chris Wood retained penalty duty, Luca Netz scored against Brest, and Callum Hudson-Odoi plus Omari Hutchinson added return-to-play evidence. Morgan Gibbs-White (15.18 xPts over GW1-3; 88.1 GW1 xMins) and Neco Williams project above 88 xMins. Hutchinson (10.86 xPts over GW1-3; 68.0 GW1 xMins) and Hudson-Odoi (9.72 xPts over GW1-3; 56.4 GW1 xMins) remain workload risks, with Ola Aina and Murillo also near 61 xMins.
• Sunderland — 7 games, 13 for, 11 against and 2 clean sheets. Brian Brobbey started centrally and scored against Rennes, Trai Hume added the winner and Enzo Le Fée supplied set pieces. Enzo Le Fée (12.93 xPts over GW1-3; 88.4 GW1 xMins), Trai Hume (11.31 xPts over GW1-3; 86.4 GW1 xMins) and Granit Xhaka all project above 86 xMins. Dan Ballard's 60.1 xMins keep him below lock status despite his defensive-contribution appeal.
• Tottenham Hotspur — 7 games, 12 for, 5 against, 3 clean sheets and no defeats. Richarlison started centrally and scored in the stadium win over Hoffenheim; Mathys Tel created both Mikey Moore goals, although Hoffenheim had changed every outfielder. James Maddison then played 45 minutes in a different XI. Richarlison (12.66 xPts over GW1-3; 74.3 GW1 xMins) is only just above the uncertainty line, while Maddison, Tel and Moore sit at 36.1, 41.9 and 18.6 xMins. The unbeaten summer does not erase a difficult 1.82 model xGA at Brentford.

Minutes and starting roles

The table shows the highest current three-Gameweek xPts projection at each club and the opening-week minutes assumption behind it. A high xPts total with low xMins is upside conditional on selection; a player near 90 xMins has a much clearer route to collecting appearance, clean-sheet and attacking-event value.

The sharpest evidence-model conflict is Arsenal's centre-forward position: Havertz started the final three matches, but the current model still assigns Gyökeres more than twice his GW1 xMins. That is a review trigger, not permission to overwrite either source. Similar uncertainty remains around Palmer and João Pedro at Chelsea, Cherki and Foden at City, Fulham's attacking full-backs and Spurs' supporting attackers.

• Strongest minutes locks: Haaland, Bruno Fernandes, Gabriel, Watkins, Thiago, Tavernier, Gibbs-White, Szoboszlai, Thiaw, Dewsbury-Hall, Ndiaye, Xhaka, Le Fée, Hume, Bassey, O'Shea and several first-choice goalkeepers and centre-backs sit above roughly 85 GW1 xMins.
• High-upside uncertainty: Gyökeres, Calafiori, Schade, Dango Ouattara, Palmer, João Pedro, Beto, Sessegnon, Rayan, Sarr, Cherki, Foden, Kerkez, Frimpong, Barnes, Hudson-Odoi, Maddison, Tel and Moore all require selection or workload confirmation.
• Friendly minutes are evidence, not a guarantee. Split squads, 70-minute training games, mass half-time changes and workload-managed competitive returns prevent raw appearance totals from being compared as if every club followed the same programme.

Injuries and availability

The most important late returns were explicit rather than inferred. Alexis Mac Allister made his first pre-season appearance on 16 August; James Maddison made his first start in almost a year and played 45 minutes; Lewis Miley returned after a fractured fibula; Callum Hudson-Odoi started and Omari Hutchinson completed a half; Manchester United reintroduced Kobbie Mainoo, Lisandro Martínez and Marcus Rashford from the bench. Those are positive workload steps, not full-fitness declarations.

Brighton described Yankuba Minteh as injured when explaining the only change from the Roma XI, and the official FPL feed continues to flag him unavailable. Manchester City's report also described workload management for players returning from the World Cup. Elsewhere, a low xMins number is not automatically an injury: it can reflect rotation, substitution risk, competition or a late return, and the model must not be used as a medical source.

• Before publication, recheck official club evidence for every player currently flagged injured, doubtful or suspended in the FPL feed; a feed label is a review queue, not an independent diagnosis.
• The opening press-conference window can supersede pre-season assumptions. Any confirmed absence, return or manager selection clue should trigger a model refresh and send the article back through fact/data review.

What changes for FPL

Build around secure minutes first, then spend risk on the roles with a clear upside case. Arsenal's attack, Brentford's front line, Gakpo, João Pedro, Hinshelwood, Davis and Le Fée deserve attention; Bournemouth, Palace, Newcastle and the Shield version of City's defence require more caution than their prices or fixtures alone might suggest.

• Prioritise locks: Haaland and Bruno Fernandes are the premium projection anchors; Gabriel, Watkins, Thiago, Tavernier, Gibbs-White and the secure promoted-club defenders offer clearer minutes foundations.
• Treat Arsenal's forwards as a live call: the model prefers Gyökeres' minutes, the final three selections preferred Havertz, and Tzolis produced the Shield assists. Wait for the manager evidence that resolves the hierarchy.
• Reward strong defensive process carefully: Brighton, Ipswich, Coventry, Liverpool, Forest and Spurs produced multiple clean sheets, but GW1 opponent strength materially changes their immediate value.
• Do not chase raw totals: Bournemouth's 22 goals include a 10-1 match; Palace's final 3-0 loss includes an extra training period; split squads and short games distort minutes and score comparisons.
• Final action: re-run xPts and xMins after the last meaningful press conferences, then move only players whose role, availability or start probability has genuinely changed.