The 2025-26 Premier League ended with Arsenal’s first title in 22 years, and then the league lost its mind. Guardiola walked away after a decade and six titles. Liverpool sacked the man who won them the 2024-25 championship. Chelsea hired the coach Real Madrid discarded after 233 days. Tottenham went from final-day survival to breaking their transfer record twice in five days. Ten of twenty clubs will start 2026-27 with a manager they didn’t have last August. Our model simulated the new season 5,000 times. The numbers below are what it believes — including the places we think it’s blind.
Arsenal at 77.7%: the number that defines our season
Arsenal won it with 85 points, 26 wins and a +44 goal difference, the title sealed on May 19 when Man City could only draw 1-1 at Bournemouth — Haaland’s stoppage-time equalizer confirming, rather than delaying, the inevitable. The model makes the champions 77.7% to repeat, 99.0% for the top four, 83 expected points, Elo 2063.8. That Elo sits almost 93 points clear of second-placed City. It’s a monstrous edge.
Is it too high? The honest case against: Arsenal have barely spent. Hincapié’s loan was made permanent for ~£34.5m, Meslier arrived free from Leeds, and the summer’s exits — Kiwior to Porto, Hein to Bremen — were pre-agreed conversions, not fresh business. Meanwhile City dropped a British-record £116m on Elliot Anderson. Elo can’t see transfer ledgers; it only sees results.
But results are exactly why 77.7% reads fair rather than inflated. The chasing pack didn’t merely fail to close the gap — it detonated itself. City lost Guardiola, Bernardo Silva (free, to Real Madrid), Akanji and Aké in one summer. Liverpool are onto their third manager in three seasons and lost Salah, Robertson and Konaté in one window. Arteta’s machine kept its spine and its manager. Standing still is a strategy when everyone around you is on fire.
The real race: three clubs, one coin
City at 82.1% for the top four are close to a lock even in transition — 12.9% for the title is the model saying post-Pep City remain elite but are no longer a repeat machine, which feels right given the outflow, even with Maresca arriving on a three-year deal (Chelsea pocketing reported £17m-plus in compensation) and Anderson’s £116m arrival. Haaland — 27 goals last season, contract to 2034 — flirted with Spain in the spring but has publicly recommitted. Treat the noise as noise.
Below them is the tightest race on the board: Aston Villa 50.6%, Man United 46.9%, Liverpool 45.3% for the top four. Three clubs, one coin flip, and the model’s reasoning differs sharply on each.
United are its biggest six-month Elo climber at +91, built on Carrick’s 11 wins in 16 after January and a third-place finish that earned him a deal through 2028. Here’s our objection: that +91 is pure momentum. As of mid-July United have sold Højlund to Napoli for ~£43m, let Casemiro go, loaned Onana out again — and signed nobody. A squad priced on a hot half-season with a thinner striker corps is exactly the profile Elo overrates. We think 46.9% is a couple of points too generous.
Liverpool cut the other way. Elo 1910.8 remembers fifth place and the chaos of Slot’s May 30 dismissal; it can’t yet see that Iraola — who just dragged Bournemouth to sixth and the club’s first European qualification in its 127-year history — is a genuine coaching upgrade, or that Isak, the £125m British-record signing, gets a manager built for him. Jacquet (£55m from Rennes) and Victor Muñoz (~£35m) patch the departures. If any of the three beats its number, it’s Liverpool.
Villa at 50.6% are the boring, correct favorite: fourth last season, Watkins’ 16 goals, and the scarcest asset in England right now — continuity. Bournemouth at 20.6% carry the residue of a departed manager; the +73 Elo surge was Iraola’s work, and Kirkland Rose inherits the squad without him or Senesi.
Tottenham: the team the model can’t price
The full arc bears repeating. Frank sacked on February 11 with Spurs 16th. Tudor’s 44-day interim: seven games, no league win. De Zerbi in on April 1 — five years, third head coach of the season. Survival secured on the final day, 1-0 over Everton, 17th on 41 points, sending West Ham down instead.
Then the summer: Mateus Fernandes for a club-record £85m from relegated West Ham, beating United to the deal after Real Madrid pulled out. The record lasted days — Tonali arrived at £92.5m up front, up to £100m, on a six-year deal. Add van Hecke at £52m, Robertson, Senesi and Dubravka on frees, Bissouma released after 111 appearances.
This is the model’s largest blind spot, and we’ll say so plainly: its rating still carries the scars of a 17th-place season and cannot see roughly £230m of new spine until it plays. Either the steepest Elo climb of 2026-27 starts in August, or this is the most expensive panic buy in Premier League history. Load Spurs into the what-if lab and watch how far the table bends when you correct for it.
The bottom: brutal to the new boys, suspicious of Sunderland
Hull, back after nine years thanks to McBurnie’s 95th-minute Wembley winner, get no romance: 99.6% relegated, Elo 1532.9. Ipswich, straight back up on 84 points, sit at 82.9% — and the model can’t know whether O’Neil, appointed June 23 after McKenna’s departure, softens or worsens that. Coventry, champions with 95 points under Lampard in their first top flight since 2000-01, are 72.5%.
Coventry is where we’d argue. A 95-point runaway champion with a settled, award-winning manager is a different animal from a play-off winner or a promoted side that just changed coach; their +54 Elo climb is still in progress. If one of the three stays up, it’s them. Sunderland at 19.9% looks strange for a team that finished seventh and reached the Europa League — until you notice Elo 1735.9 is still bottom-six material and Thursday football is a tax on thin squads.
Movers and wildcards
The movers board tells the summer’s story in one line: United +91, Bournemouth +73, Sunderland +56, Forest +55, Coventry +54 — and Chelsea -52. Forest made the sharpest hire of the window: Glasner, fresh from winning the Conference League with Palace and only the third coach ever to win both it and the Europa League, appointed July 6. Palace replaced him with Sage from Lens. Chelsea’s -52 is the ugliest number on the site — Maresca walked out on New Year’s Day with the club fifth, and Alonso arrives off his 233-day Madrid flameout, with Cucurella sold to Real for up to £51.8m and Palestra (up to £47m) his first signing. Two decorated coaches, one damaged rating.
And two stories stay open: Salah remains a free agent after Egypt’s World Cup exit, Saudi clubs circling, nothing confirmed; relegated West Ham reportedly need to raise ~£100m beyond the Fernandes sale, with the Bowen links to Chelsea, Liverpool and United still rumor — he’s contracted to 2030 and hinting he stays.
What we’ll be graded on
Three claims, in public: Arsenal at 77.7%, Hull at 99.6%, and the Villa–United–Liverpool coin flips at 50.6/46.9/45.3. We’ve told you where we’d overrule the machine — Liverpool up, United down, Coventry alive, Spurs unpriceable. The full table, week by week, lives at the full Premier League forecast, and every call gets scored on the track record. The season starts in August. The grading already has.
