Matchweek 1, previewed

Coventry City have not played a game in England's top division since May 2001. In the years since they have been relegated three more times, gone into administration, groundshared in Northampton and Birmingham, and spent a season where their own supporters were not sure the club would still exist at the end of it. On Friday night they walk out at the Emirates while the champions of England unfurl a banner. Football does not usually do symmetry this neat.

The Summer

Arsenal have spent theirs solving a problem most clubs would love to have: how to improve a title-winning midfield. Bruno Guimaraes arrives from Newcastle for EUR87.5m to sit alongside Declan Rice and Martin Zubimendi, which is less a midfield than a toll booth. Piero Hincapie came in for EUR40m and Christos Tzolis for the same, with Illan Meslier arriving free. EUR168m out, EUR46m back through Leandro Trossard and Jakub Kiwior. Mikel Arteta is chasing something no Arsenal manager has managed since 1934-35, a successful title defence, and only three clubs have retained the Premier League at all: United under Ferguson, Chelsea under Mourinho, City under Guardiola.

Coventry have not come up quietly. Roughly EUR108m has gone out on Caleb Yirenkyi (EUR27m), goalkeeper Carl Rushworth from Brighton (EUR25.7m) and Loum Tchaouna (EUR23m), which is a serious outlay for a promoted club and a sign they intend to compete rather than survive on sentiment. The problem is arithmetic. Arsenal's squad is valued around EUR1.41bn. Coventry's is EUR265m, nineteenth of twenty. That is not a gap you close with one window.

Who To Watch

Arsenal's returning numbers are led by Kai Havertz, whose 0.76 non-penalty expected goals per 90 last season was the best at the club, though across only 582 minutes. Gabriel Martinelli (0.47 npxG, 0.23 xA per 90 across 1,025 minutes) and Mikel Merino (0.47 and 0.21) both contributed heavily, and Bukayo Saka remains the most complete of them: 0.32 npxG and 0.34 xA per 90 across 2,239 minutes, comfortably the biggest workload of any Arsenal attacker. The genuine worry is William Saliba's back, injured at the World Cup. Arsenal's win rate since his 2022 debut drops from 68.7% with him to 47.6% without.

For Coventry, we hold no Premier League production data on a single player, because none of them have any. That is worth stating plainly rather than filling with adjectives.

Form And History

Arsenal closed last season W5 L1 from six, conceding once in five of them, and their 27 goals against across the campaign was built on an average 0.74 expected goals conceded per game, the fourth lowest on record since 2012-13. These clubs have no head to head on our books at all. The last time they met competitively, most of this Arsenal squad was not born.

The Numbers

Our model makes Arsenal 81%, the draw 15%, Coventry 4%, with a projected 2.49 to 0.47. The single likeliest scoreline is 2-0 at 16%, then 3-0 at 13% and 1-0 at 12%. Over 2.5 goals sits at 57% and both teams to score at just 35%, the second lowest on the card.

The Underlying Numbers

Arsenal's title was built at the back and the profile is stark. They kept a clean sheet in 50% of matches last season, the best in the division, conceding 0.71 goals a game and facing only 2.4 shots on target per match. No other side faced fewer than 3.3. They also failed to score in just 8% of games, the lowest rate in the league, so the route to an upset here is not keeping them out for 90 minutes; it is scoring first and asking questions of a side that rarely has to chase. Arsenal averaged 14.5 shots and 5.7 corners a game, and Mikel Arteta's team have leaned on set pieces more than any recent Arsenal side.

The Read

Everything points one way and the only honest caveat is the calendar. This is Matchweek 1. Nobody has kicked a competitive ball, Coventry are unknowable by definition, and promoted sides have embarrassed the arithmetic before. Sunderland were written off by everyone including us last August and finished seventh. We are not expecting a repeat on Friday. We are also not pretending our 4% is anything other than a model's opinion about a team it has never seen.

Tottenham have spent EUR267m this summer. Sandro Tonali for EUR108m, Mateus Fernandes for EUR99m, Jan Paul van Hecke for EUR60m. They have also finished seventeenth in each of the last two seasons. Both of those sentences are true, and reconciling them is the most interesting question of the opening weekend.

The Summer

Roberto De Zerbi gets his first full season, and gets it without European football, which matters more than it sounds. In the last four seasons Newcastle made the top five on both occasions they had no European commitments and missed out both times they did; the pattern is not unique to them. Cristian Romero left for EUR40m and Luka Vuskovic for EUR54m, so this is a genuine rebuild rather than a top-up. EUR164m recouped against EUR267m spent.

Brentford's summer was quieter and arguably cleverer. EUR84m spent, only EUR7m recouped, Mamadou Sangare the headline arrival at EUR48m with Jannik Schuster (EUR18m) and Jaidon Anthony (EUR17.5m) behind him. Keith Andrews continues. They finished ninth, and our model has them ninth again, which after the last few years counts as an institution.

Who To Watch

Brentford's threat comes from wide. Dango Ouattara returns with 0.34 npxG and 0.23 xA per 90 across 2,303 minutes, and Kevin Schade played 2,781 minutes at 0.36 npxG. Mikkel Damsgaard is the creator, 0.28 xA per 90, and Vitaly Janelt chips in 0.24 from deeper.

Spurs' returning production is thinner than EUR267m suggests, which is rather the point. Richarlison leads at 0.46 npxG per 90 across 1,926 minutes, with Mathys Tel and Wilson Odobert both at 0.42 combined in limited minutes, and Dominic Solanke at 0.38 across only 989. The new midfield of Tonali and Fernandes has no Premier League record with this club at all, so what happens on Saturday is genuinely new information.

Form And History

Spurs actually finished last season reasonably, W3 D2 L1, which sits oddly against seventeenth place and tells you how bad the middle of it was. Brentford drew four of their last six. The head to head has been one-sided: Tottenham have won four of the last five on record, including 2-0 and 3-1, though the most recent meeting at the Gtech in January finished 0-0.

The Numbers

Brentford 43%, draw 29%, Spurs 28%, projected 1.51 to 1.18, 1-1 likeliest at 14%. Both teams to score sits at 56%, the second highest on the card, and over 2.5 goals at 50%.

The Underlying Numbers

Two sides who found scoring hard. Brentford took 10.7 shots a game last season, second-lowest in the division, and failed to score in 32% of matches. Tottenham managed 11.5 shots and blanked in 18%. What separates them is at the back: Spurs conceded 1.50 goals a game, the worst of any side still in this league, and kept clean sheets in 24%. Brentford conceded 1.37. There is also a discipline story worth watching. Tottenham averaged 2.89 card points a game, joint-worst in the division with Chelsea, against a two-side league average of 3.99. If the referee is one of the stricter appointments, and our records have J Brooks at 4.7 card points a game and D Coote at 4.5 against M Atkinson's 2.6, that matters more for Spurs than for most.

The Read

Our model is materially lower on Spurs than the wider market, which has them around seventh; we have them tenth. The reason is worth stating plainly rather than dressing up: our model reads results, and Spurs' results have been dreadful for two years. It cannot see EUR267m, it cannot see De Zerbi, and it cannot see an empty midweek calendar. That is a known blind spot rather than a considered opinion. Saturday is a first look at which of those things matters more.

On paper this is the least glamorous fixture of the weekend: thirteenth against fifteenth, four points between them last season, no European glamour on the day. It is also the game where the summer's business is most likely to decide something, because Everton and Crystal Palace essentially did their business with each other.

The Summer

Dwight McNeil went from Goodison to Selhurst Park for EUR25.8m. Brennan Johnson came the other way for precisely the same fee. Everton spent about EUR99m in total, adding Merlin Rohl (EUR25m) and Tyrique George (EUR21m), and recouped EUR26m. Palace went the other way entirely, cashing in: Maxence Lacroix to Chelsea for EUR60.7m, EUR93m recouped against EUR50m spent, a net profit of EUR43m.

Palace also have a new manager and a European hangover to manage. Oliver Glasner left for Nottingham Forest after winning the Conference League, and Pierre Sage inherits a side that now has Thursday nights to negotiate. A first sustained European campaign plus a new voice in the dressing room has undone better squads than this one. Everton, by contrast, are one of very few clubs in the division with the same manager, no European football and a quiet, tidy window, which in a summer of nine managerial changes counts for something.

Who To Watch

Everton's threat is centre-forward shaped. Beto returns with 0.64 npxG per 90 across 1,552 minutes, the highest at the club, with Thierno Barry at 0.43 across 1,921. The creative burden sits with Kiernan Dewsbury-Hall (0.22 xA per 90 across 2,649 minutes) and Iliman Ndiaye, whose 2,805 minutes were the most of any Everton attacker.

Palace lean on Jean-Philippe Mateta, 0.52 npxG per 90 across 2,225 minutes and comfortably their most reliable source of goals, with Adam Wharton the metronome behind him at 0.23 xA per 90 across 2,596. Daniel Munoz remains one of the most productive attacking full-backs in the division, 0.19 xA per 90 from defence.

Form And History

Neither ended last season well. Everton took two points from their final six, Palace one from theirs. The head to head is genuinely tight and genuinely low-scoring: five meetings on record, Everton winning three, two draws, and only once has either side scored more than twice. The most recent was a 2-2 at Selhurst Park in May.

The Numbers

Everton 42%, draw 31%, Palace 27%, projected 1.34 to 1.04. The likeliest score is 1-1 at 15%, then 0-0 and 1-0 at 11% each. Over 2.5 goals sits at just 42%, and the draw is doing a great deal of work in that distribution.

The Underlying Numbers

This is a game between two of the least productive attacks in the division. Palace scored 1.08 goals a game last season, the lowest of any side still in the league, and failed to score in 32% of matches. Everton were not much better at 1.24 and 26%. Neither creates volume: Everton managed 11.1 shots and 4.3 corners a game, Palace 11.6 and 4.2, both bottom five. Both defences faced 4.3 shots on target a game, identical. Our 42% for over 2.5 goals is not pessimism, it is description. Palace pick up 2.11 card points a game against a league average of 3.99 across both sides, so this should be scrappy rather than bad-tempered.

The Read

We rate these two almost level for the season, thirteenth and fourteenth, and home advantage is most of what separates them here. Our model rates Everton slightly higher than the market implies, largely because it rewards stability in a summer where almost everyone else changed something significant. The counter is obvious: Palace won a European trophy four months ago and Everton finished thirteenth.

Hull City are back in the Premier League for the first time since 2017, and the fixture computer has handed them Manchester United at home in the Saturday lunchtime slot. The MKM Stadium will be at its loudest in a decade. Our model thinks it will not be enough, and it is not close about it.

The Summer

Hull's promotion was the strangest story of the Championship season. They finished sixth, went up through the play-offs, and beat Middlesbrough 1-0 at Wembley through Oli McBurnie in the fifth minute of second-half stoppage time. Their underlying numbers trailed their results all year. Sergej Jakirovic has responded by rebuilding almost everything: nineteen players in, sixteen out, EUR89m spent on Nobel Mendy (EUR23.4m), goalkeeper Konstantinos Tzolakis (EUR23.3m) and Lucas Herrington (EUR14.8m). A squad valued at EUR144m remains the smallest in the division by a distance, roughly a tenth of United's EUR874m.

United arrive in a better place than at any point in a decade. Michael Carrick took over on 12 January and won 39 points from the final 17 games, more than any side in the division including the champions, three more than Arsenal. Andrey Santos came in for EUR56.3m and Youri Tielemans for EUR41m from Aston Villa, with Rasmus Hojlund leaving for EUR44m. Third place last season, Champions League football to juggle now.

Who To Watch

United's returning production is deep. Joshua Zirkzee posted 0.60 npxG per 90, Benjamin Sesko 0.69 across a far larger 1,636 minutes, and Bryan Mbeumo 0.45 npxG with 0.17 xA across 2,676. The engine remains Bruno Fernandes, whose 0.52 expected assists per 90 across 3,082 minutes, essentially a full season, is the highest creative output at the club by some way. If Hull sit deep, he is the player who unpicks it.

Hull, like Coventry, carry no Premier League production data because nobody in the squad has any. Tzolakis in goal may be the most consequential signing of their summer for exactly that reason.

Form And History

United closed the season W5 D1 from six, including 3-2 at home to Liverpool and 3-0 away at Brighton. No head to head on record between these clubs; Hull's last top-flight season ended before Carrick had taken a coaching badge at United.

The Numbers

United 70%, draw 22%, Hull 9%, projected 0.59 to 2.01. The likeliest score is 0-2 at 15%, then 0-1 at 14% and 0-3 at 10%. Over 2.5 goals is 48%, both teams to score 40%. A second public projection we track is even harsher, putting United at 3.2 goals to our 2.01.

The Underlying Numbers

United's attacking profile last season was the best in the division on volume: 15.7 shots per game, level with Manchester City, and 5.7 shots on target, more than anybody. They failed to score in only 11% of matches. The weakness is at the other end, where 1.32 goals conceded and 21% clean sheets are mid-table numbers dressed up by a third-place finish. That is the crack Hull have to find, and their own record offers no evidence either way because none of it exists at this level. United also took 4.8 corners a game, among the lowest of the top sides, which tells you they score from open play rather than dead balls.

The Read

We rate Hull the weakest side in the division and give them a 99% chance of being involved in a relegation fight at some stage. That is a hard thing to say about a club nobody has seen at this level, and it rests on a squad value and a promotion route rather than on evidence from this division. Treat the opening weekend as evidence gathering, not confirmation.

Ipswich Town went down, came straight back up, and open at home against the club that did the single most surprising thing in the Premier League last season. Sunderland were promoted, were written off by everybody including us, and finished seventh. That is the tension running through this fixture, and it is why our model's read on it feels uncomfortable to write down.

The Summer

Sunderland's reward for seventh was Europa League football and a summer of near-total inactivity: EUR0m spent. Not a modest amount. Zero. Thomas Meunier arrived on a free, Simon Adingra and Luis Semedo returned from loans, and Eliezer Mayenda left for EUR22m with Anthony Patterson going for EUR9.35m. A club that overachieved spectacularly has chosen to stand still while adding Thursday nights to the workload. Opta's own read on them is bleak for the same reason: only Burnley and Wolves managed a lower open-play expected goals total last season, and Sunderland faced the third-most shots in the division.

Ipswich have done the opposite. Gary O'Neil has replaced Kieran McKenna, and roughly EUR128m has gone out on Emersonn (EUR28m), Abdul Fatawu (EUR23.45m) and Florentino (EUR18.7m). They finished second in the Championship on 84 points and bounced back at the first attempt, which is rarer than it sounds.

Who To Watch

Sunderland keep the pieces that got them there. Wilson Isidor posted 0.44 npxG per 90 across 1,142 minutes and Brian Brobbey 0.33 across 1,943, while Enzo Le Fee played 2,953 minutes, essentially every one available, and created at 0.23 xA per 90. Chris Rigg, still a teenager, contributed 0.23 combined per 90 in limited minutes and is the one most likely to make this season look different.

Ipswich carry no Premier League production on record for the current squad. Their relegation season is not in the model's read on them either, because we treat bounce-back clubs as promoted rather than trusting form from a campaign they were relegated in.

Form And History

Ipswich finished their last Premier League season badly: five defeats and a draw from their last six, including 0-4 at home to Arsenal. Sunderland closed strongly, beating Everton 3-1 and Chelsea 2-1 in their final two. No head to head on record at this level.

The Numbers

Ipswich 29%, draw 34%, Sunderland 37%, projected 0.95 to 1.10. This is the tightest game of the weekend and the draw is the single likeliest outcome, with 1-1 and 0-0 both at 15%. Over 2.5 goals sits at 34%, the lowest figure on the entire card.

The Underlying Numbers

Here is the case against Sunderland in one line: last season they managed 10.4 shots and 3.3 shots on target a game, both the lowest of any side still in this division, from a league-low 3.7 corners. They failed to score in 34% of matches. A team finishing seventh on those numbers is a team riding something that does not usually last, which is precisely why our model expects regression and why Opta flagged them too. The counterweight is that they conceded only 1.26 a game and kept clean sheets in 29%, better than Chelsea, Newcastle and Manchester United managed. They were hard to beat rather than lucky, which is a meaningfully different thing.

The Read

Our model has Sunderland sixteenth, expecting regression from seventh, and Ipswich eighteenth. So it reads this as a relegation six-pointer in August, involving a side that finished seventh four months ago. We should be honest about the track record here: we called Sunderland wrong last season, badly, and so did almost everyone. Their underlying numbers say the model is right. Their actual results say be careful.

Nottingham Forest sold Elliot Anderson to Manchester City for EUR135m. That is the headline of their summer and very nearly the whole of it: EUR144m out, EUR60m back in, a net profit of EUR84m and a squad that is, on paper, weaker than the one that finished sixteenth. Our model has them eighth. That deserves explaining rather than asserting.

The Summer

Ousmane Diomande arrived from Sporting for EUR40m and Gustavo Sa for EUR20m, with Steven Benda in on a free. The more significant arrival wears a suit. Oliver Glasner has left Crystal Palace, where he won the Conference League, and taken over at the City Ground. He is the sort of appointment that tends to move a team's ceiling quickly, and our model's optimism is partly that and partly that Forest's underlying numbers ran ahead of their sixteenth-place finish.

Leeds spent EUR84m, most of it on goalkeeper James Trafford from Manchester City for EUR46.7m, with Tarik Muharemovic at EUR37m. They also took Harry Wilson from Fulham on a free, a piece of business that at least one respected pundit considered significant enough to relegate Fulham over. Same manager, no European football, fourteenth last season.

Who To Watch

Forest's creative axis survives the Anderson sale. Morgan Gibbs-White played 3,142 minutes, more than anyone at the club, contributing 0.31 npxG and 0.13 xA per 90. Omari Hutchinson is the sharper creator at 0.27 xA per 90, and Chris Wood remains a genuine penalty-box presence at 0.33 npxG per 90. Callum Hudson-Odoi adds 0.33 combined across 1,857 minutes.

Leeds are led by Lukas Nmecha (0.49 npxG per 90) and Dominic Calvert-Lewin, whose 2,732 minutes at 0.49 npxG per 90 made him their most-used and most productive forward. Noah Okafor at 0.36 and Brenden Aaronson at 0.41 combined give the front line more depth than most promoted-era Leeds sides have had.

Form And History

Forest ended the season W3 D2 L1, including a 5-0 win at Sunderland and 3-1 at Chelsea, which is precisely the sort of run that makes a model like ours optimistic. Leeds were solid but unspectacular, W3 D2 L1. The head to head is lively: four meetings on record, three of them won by three or more goals by one side, and they split last season's fixtures 3-1 each way.

The Numbers

Forest 48%, draw 29%, Leeds 23%, projected 1.50 to 0.98, with 1-1 the likeliest score at 14%. Over 2.5 goals sits at 45%.

The Underlying Numbers

And here is the case against our own optimism. Forest failed to score in 37% of matches last season, the worst rate of any side still in this division, from 12.5 shots a game. A team that blanks in more than one game in three is not obviously a top-eight team, and Elliot Anderson, who has gone, was central to the creation that did exist. In Forest's favour: 1.34 goals conceded and a disciplined 1.58 card points a game, and Glasner's Palace side were built on exactly that defensive base before the goals arrived. Leeds conceded 1.47 a game, the worst figure in this fixture, and kept clean sheets in only 21%.

The Read

This is where our model sits furthest from everyone else. The wider consensus has Forest around thirteenth. We have them eighth, a six-place gap that is the largest disagreement on our board, though it is still inside our own measured error of roughly three places. If Glasner needs a season to bed his ideas in, as he did at Palace, we will look silly by October. Watch the press. It is the fastest-arriving part of a Glasner side.

Aston Villa finished fourth and won the Europa League, beating Freiburg 3-0 in the final. Then they sold Morgan Rogers to Chelsea for EUR138m, a British record, and Youri Tielemans to Manchester United for EUR41m. EUR234m out of the door in one summer, EUR106m back in. Unai Emery remains, which is the single most reassuring fact in that paragraph.

The Summer

Johan Manzambi arrived for EUR60m and Joao Gomes for EUR40m, both interesting, neither obviously a Rogers replacement. Villa now have Champions League football to juggle alongside a squad missing its best creator and a first-choice midfielder. Our model has them twelfth, down from fourth, and it is worth being upfront that this is among our least popular calls: the wider consensus has them around sixth.

Brighton, as usual, have run the market rather than been run by it. Jan Paul van Hecke went to Spurs for EUR60m and Luka Vuskovic arrived from the same club for EUR54m, a pleasing bit of circularity. Zadok Yohanna came in for EUR28m and Pascal Struijk from Leeds for EUR23.2m, with Carl Rushworth sold to Coventry for EUR25.7m. Same manager, eighth last season, and a Conference League play-off round to negotiate rather than a full campaign.

Who To Watch

Villa still have Ollie Watkins, and he is still the point: 0.58 npxG per 90 across 2,908 minutes, more than anyone at the club and the highest workload of any forward in this fixture. Behind him the creation is spread thin. Ross Barkley managed 0.43 combined per 90 but across only 818 minutes, Emiliano Buendia 0.38 across 1,778, and John McGinn 0.20 xA per 90 from 2,221. Losing Rogers takes a large slice out of exactly that band.

Brighton's spread is the opposite: nobody dominant, everybody contributing. Kaoru Mitoma leads at 0.43 combined per 90, with Maxim De Cuyper posting a striking 0.30 xA per 90 from left-back across 1,412 minutes. Yankuba Minteh played 2,410 minutes at 0.40 combined, and Georginio Rutter adds 0.23 xA per 90.

Form And History

Both finished erratically. Villa won three of their last six including 4-2 over Liverpool, but also lost at Fulham. Brighton lost three of six including 0-3 at home to Manchester United. The head to head is a shootout: five meetings on record, four decided by a single goal or fewer except a 3-4 at the Amex in December that flattered nobody's defence.

The Numbers

Brighton 44%, draw 29%, Villa 26%, projected 1.48 to 1.10, 1-1 likeliest at 14%. Both teams to score 53%, over 2.5 goals 48%.

The Underlying Numbers

Villa's numbers last season were better than their post-transfer squad suggests they will be again: 12.7 shots and 4.5 on target a game, 1.47 goals scored. The concern is defensive consistency, 4.4 shots on target faced per game and clean sheets in only 24%, the joint-lowest in this fixture. Brighton were the more balanced side, 13.0 shots and 4.6 on target, conceding 1.21, the better figure of the two. Both are among the more carded teams in the division, Brighton 2.26 and Villa 1.61 card points a game, and the head to head has produced 3-4 and 3-0 scorelines in the last two seasons. Our 53% for both teams to score reflects a fixture that has rarely been quiet.

The Read

Our model rates Brighton higher than the market implies and Villa considerably lower, and this fixture is where those two views collide most directly. The honest framing: we are pricing a squad that lost EUR234m of talent against one that has recycled successfully for a decade. The counter is that Emery is one of the best coaches in the division and has absorbed departures before. Ask again in November.

Pep Guardiola has left Manchester City. It is still strange to type. Enzo Maresca has the job, and Sunday afternoon is the first competitive team sheet at the Etihad in a decade that somebody else has written.

The Summer

Maresca arrives from eighteen months at Chelsea that read differently depending which number you lead with. He won Champions League qualification, the Club World Cup and the Conference League. He also averaged 1.74 points per game, the worst of anyone to manage Chelsea for at least 30 Premier League games this century. He inherits a squad built for his possession game and fronted by Erling Haaland, a three-time Golden Boot winner who can now match the record four held jointly by Thierry Henry and Mohamed Salah. City spent EUR177m, EUR135m of it on Elliot Anderson from Nottingham Forest, and sold James Trafford to Leeds for EUR46.7m and Manuel Akanji for EUR15m. Rodri continues to be linked with Barcelona; Bernardo Silva has already gone to Real Madrid.

Bournemouth are the quietly excellent story of the last two seasons: sixth last year, Europa League football now, and a new manager in Marco Rose after Andoni Iraola left for Anfield. About EUR80m spent, Antonio Silva the headline arrival at EUR25m. Our model likes them a good deal more than most, sixth again this season against a consensus of around tenth.

Who To Watch

Bournemouth's returning production is genuinely good for a club of their size. Evanilson posted 0.38 npxG per 90 across 2,810 minutes, Marcus Tavernier 0.46 combined across 2,779, and Justin Kluivert the best rate of all at 0.44 combined, though across only 934 minutes. That is a front line that already works, which is the argument for Rose picking up where Iraola left off rather than starting again.

Form And History

Bournemouth were unbeaten in their last six, W3 D3, including 1-1 at home to City in May and a 2-1 win at Newcastle. That draw matters here: the head to head over five meetings is closer than the table suggests, with Bournemouth winning 2-1 at the Vitality in November 2024 and drawing the most recent.

The Numbers

City 65%, draw 22%, Bournemouth 13%, projected 2.14 to 0.88. That is the highest projected goal total on the card, 58% for over 2.5, with 2-0 likeliest at 11% and both 1-1 and 2-1 at 10%. Notably, a second public projection we track has City at only 1.6 goals against our 2.14, so we are the aggressive ones here.

The Underlying Numbers

City's underlying profile last season was still elite even in a title-losing campaign: 15.7 shots a game, 5.4 on target, 6.5 corners, the highest volume in the division, and 2.03 goals a game. They conceded 0.92 and kept clean sheets in 42%, second only to Arsenal. Bournemouth are no soft touch, 13.8 shots and 4.7 on target a game, more than Chelsea, Newcastle or Tottenham managed, and they failed to score in only 18% of matches. Their weakness is discipline: 2.47 card points a game, among the highest in the league, and against a side that averages 6.5 corners, giving away free kicks in dangerous areas is the fastest way to lose this.

The Read

Our model rates Bournemouth higher than most and this is a reasonable first test of that. City are the strongest side we project on a per-match basis, but the first competitive game under a new manager is precisely where teams are least themselves, and Maresca's Chelsea were often slow starters. Bournemouth's Thursday nights have not begun yet, which means their legs are as fresh as they will be all season.

The best game of the weekend, and the one where two clubs have taken opposite approaches to being taken apart. Between them Newcastle and Liverpool have lost their manager, their best midfielder, their captain's-armband forward and the greatest goalscorer in either club's Premier League history. Both are starting again on the same afternoon.

The Summer

Newcastle lost Eddie Howe at the end of July, then sold Sandro Tonali to Tottenham for EUR108m and Bruno Guimaraes to Arsenal for EUR87.5m. EUR266m out. Matthias Jaissle takes over with EUR161m reinvested in Bazoumana Toure (EUR50m), Aladji Bamba (EUR35.5m) and Lukas Hornicek (EUR30.6m). Whether that reads as a rebuild or an evacuation depends entirely on how quickly Jaissle's ideas land. One genuinely encouraging note: in the last four seasons Newcastle have reached the top five on both occasions they had no European football and missed out both times they did. They have none this year.

Liverpool finished fifth, which at Anfield counted as a crisis, and Arne Slot left. Andoni Iraola arrives from Bournemouth. Mohamed Salah has gone after nine years as the club's all-time leading Premier League scorer (191) and assister (93), and nobody has directly replaced him. EUR104m spent on Jeremy Jacquet (EUR63.6m) and Victor Munoz (EUR40m), with Ronald Araujo on loan to partner Virgil van Dijk.

Who To Watch

This is where Liverpool's case sits. Hugo Ekitike returns at 0.58 npxG and 0.12 xA per 90 across 1,835 minutes, the best combined rate at the club. Cody Gakpo played 2,758 minutes at 0.54 combined. Alexander Isak, restricted to 717 minutes by injury and form last season and scoring only four all year, still posted 0.41 npxG per 90, and if he and Florian Wirtz take the second-season step the club is paying for, this front line is a different proposition. Rio Ngumoha's 0.37 xA per 90 across 515 minutes is the most intriguing small sample at the club.

Newcastle counter with William Osula (0.57 npxG per 90 across 777 minutes), Harvey Barnes (0.52 combined across 1,947) and Nick Woltemade (0.48 across 1,953). Jacob Murphy remains the creator at 0.26 xA per 90.

Form And History

Newcastle were inconsistent to the end, W2 D1 L3 from six. Liverpool were worse, W2 D2 L2, including 4-2 at Villa. The head to head is reliably chaotic: five meetings on record producing 4-2, 3-3, 2-0, 3-2 and 4-1, every one of them with at least three goals and four of the five won by Liverpool.

The Numbers

Newcastle 33%, draw 28%, Liverpool 39%, projected 1.40 to 1.54. Both teams to score at 61% is the highest on the card and over 2.5 goals at 56% is second. The likeliest score is 1-1 at 13%, then 1-2 at 9%.

The Underlying Numbers

Liverpool generated more than their fifth-place finish suggests: 15.5 shots a game, third in the division, 6.1 corners, and they failed to score in only 11% of matches. What went wrong was at the back, 1.39 conceded and clean sheets in just 26%. Newcastle were the mirror image of nobody in particular, 12.9 shots, 4.6 on target, but conceded 1.45 a game and kept only 21% clean sheets, the joint-worst in this fixture. Two attacks that function and two defences that do not is the clearest reason our model puts both teams to score at 61%, the highest figure of the weekend, and it is exactly what the head to head has delivered: 4-2, 3-3, 2-0, 3-2, 4-1.

The Read

A genuine coin flip between two sides mid-rebuild, and our model leans only narrowly to Liverpool. Neither squad has played a competitive game under its new manager, both have lost more talent than they have replaced, and the head to head suggests defensive certainty will not feature. If you want one game this weekend that tells you something real about the season, it is this one.

Chelsea spent EUR408m this summer. Not a typo, and comfortably the largest outlay in the division. Eighteen players in. Monday night at Craven Cottage is the first time anyone gets to see what Xabi Alonso does with them.

The Summer

Morgan Rogers arrived from Aston Villa for EUR138m, a British record, with Maxence Lacroix (EUR60.7m) from Crystal Palace, Marco Palestra (EUR57m) and Geovany Quenda (EUR50m) behind him. Andrey Santos left for EUR56.3m and Marc Cucurella for EUR55m, so EUR162m came back. Alonso has averaged 2.2 points per game across the Bundesliga and La Liga since 2022-23, at Leverkusen and Real Madrid, and he gets to arrange all of this with no European football whatsoever. That combination, elite coach plus enormous squad plus a free midweek, is the case for Chelsea as the season's most interesting side. Our model buys it: fourth, up from tenth.

Fulham are the opposite kind of story. Marco Silva never finished lower than thirteenth in four seasons and has gone. Alvaro Arbeloa replaces him, drawing on Real Madrid contacts for Gonzalo Garcia (EUR40m) and Cesar Palacios, with Shea Charles in for EUR30.4m. Garcia is genuinely exciting and also has nine La Liga starts to his name. Harry Wilson left for Leeds on a free. Our model has Fulham seventeenth, and we are not alone in the worry: this is the one established club a couple of respected pundits have going down.

Who To Watch

Fulham's returning production is modest and top-heavy. Rodrigo Muniz leads at 0.49 combined per 90, but across only 1,001 minutes. Emile Smith-Rowe managed 0.26 across 1,933 and Alex Iwobi 0.22 across 2,473, which is a lot of minutes for a fairly thin return. If Garcia does not settle quickly, the goals are a genuine problem.

Chelsea's is the opposite. Estevao posted 0.41 npxG and 0.30 xA per 90 across 854 minutes, the best rate at the club. Joao Pedro played 2,698 minutes at 0.53 npxG per 90. Enzo Fernandez played 3,149 minutes, essentially everything, at 0.55 combined. Pedro Neto adds 0.27 xA per 90. That is a functioning attack before Rogers has kicked a ball.

Form And History

Chelsea finished dreadfully: one win in six, including 0-1 at home to United and 1-3 to Forest. Fulham were better, W2 D2 L2, beating Newcastle 2-0. The head to head is genuinely close, five meetings split 3-2 to Chelsea with four decided by a single goal, and Fulham won the most recent 2-1 in January.

The Numbers

Chelsea 46%, draw 29%, Fulham 25%, projected 1.06 to 1.50, 1-1 likeliest at 14%. Over 2.5 goals 47%.

One honesty note: the odds provider has not posted this fixture, so unlike every other game this weekend we have no market to compare ourselves against. We are showing our number alone rather than implying a comparison we cannot make.

The Underlying Numbers

Fulham's problem is precision rather than volume. They took 12.7 shots a game last season, mid-table, but only 3.6 hit the target, the second-lowest conversion of shots into shots on target in the division, and they failed to score in 29% of matches. Chelsea managed 13.4 shots and 4.3 on target with 6.1 corners a game, joint-third most in the league, and blanked in only 18%. The wildcard is temperament. Chelsea averaged 2.89 card points a game last season, joint-worst in the division alongside Tottenham, and a new manager with eighteen new players in a London derby on a Monday night is not the setting where that improves. If the appointment is one of the stricter referees on our records, that is a live factor.

The Read

The squad-value gap is roughly EUR1bn. Chelsea should win this and our model says so. The hesitation is that Chelsea's actual results last season were bad, one win in their final six, and eighteen new players take time to become a team. Monday night, last game of the weekend, and Fulham's season may be defined by how quickly Arbeloa's ideas take hold.