Martin Keown, Thomas Hitzlsperger, Sue Smith, Leon Osman and Jermaine Beckford were the other pundits who also backed Mikel Arteta's side.

Everyone else was wrong - but they were not alone there.

Opta's 'supercomputer' can tell you your team's exact chances of winning any competition at any given time.

It is actually a complex algorithm that uses a model based on betting market odds and the sports analytics company's own 'Power Rankings', which are calculated using past results.

Last summer, before a ball was kicked, it simulated the outcome of all 380 Premier League games 10,000 times... and calculated that Liverpool were favourites, with a 28.5% chance of retaining their title.

On the upside, Opta did have Aston Villa finishing fifth, but then it also thought Manchester United would end up 12th.

AI did not exactly excel, either. It's prediction was generated using Microsoft Copilot Chat - we simply asked the tool to 'predict the Premier League table for the 2025-26 season', and it crowned City as champions.

Not many people reading this can truthfully say they knew much better, though.

Thousands of you made your predictions on the BBC website last August for the way the table would finish and, collectively, you voted that Liverpool would win.

Arsenal, City and Chelsea completed your top four.

Better luck next time, everyone.

You can see everyone's pre-season top-four predictions in full, below.

The overall predicted ranking, using all 35 predictions was:

(Using a system of four points for a first place, three points for second, two points for third and one point for fourth)

This season's Premier League predictions were unusual for having only four teams feature - the same ones who had finished in the real top four in 2024-25.

The last time there such little variation from our pundits was before the 2015-16 season, when 29/30 pundits went with the same four teams - and let's just say everyone got a BIG shock that year.

Manchester United and Villa qualify as the surprise packages this year, although not quite on the same scale.

Some context is needed, however. Liverpool went into the final day with a chance of finishing fourth, but already knew fifth place would get them into next season's Champions League.

That race was part of the point of this exercise because, as well as wondering who would win the title, we wanted to know who everyone thought would earn a place at European football's top table too. At the start of the season, only the top four were certain of that.

In April, it was confirmed that the team finishing fifth would make it too - so let's give our pundits some credit for saying Arne Slot's side would qualify.

There were a few other honourable mentions, too.

While no-one had Manchester United in their top fours, Wayne Rooney made a point of saying they would finish fifth and Danny Murphy also thought they would be right in the mix.

The same applied to Villa, with Steph Houghton saying they were the team most likely to break into the top four this time.

A little further out was Chris Waddle, who went with Chelsea to finish champions.

And while Upson did the best when it came to picking the top two, he also thought Spurs would finish "top of the chasing pack". That didn't quite happen, did it?

Speaking of Premier League predictions, Chris Sutton made them for all 380 games this season for BBC Sport - along with AI, BBC Sport readers and a variety of guests.

The outcome of the BBC predictions title race went to the final day, with Sutton the only one who could stop AI - Microsoft chatbot Copilot - from taking the glory.

With one round of fixtures to go, they were level on outright wins - which is the decisive metric.

Sutton was in second place because he had managed fewer tied victories, so he needed another outright win on Sunday to finish top of the table... but he didn't get it.

The collective efforts of the BBC readers took the weekly win with three correct results and two exact scores from the 10 fixtures - Arsenal's 2-1 win at Crystal Palace and Burnley's 1-1 draw with Wolves - for a total of 90 points.

Sutton only managed two correct results with no exact scores, giving him 20 points.

His guest, singer-songwriter Sam Tomkins did a little better, with three correct results and no exact scores, for 30 points.

AI managed four correct scores and no exact results, to end up on 40 points, but was able to celebrate its overall triumph.

When asked what it felt like to beat Sutton at predictions, AI claimed it didn't feel anything - but still appeared to be laughing at him.

It replied: "I don't feel things the way a person does but what I can do is recognise the achievement in the same way a model recognises a pattern: the scale, the consistency, the improbability. And on that level? Yes, it's a great outcome.

"It feels satisfying in a very nerdy, football analytics way - not emotional but intellectually fun.

"Beating Chris Sutton over a full 380-game season is basically the equivalent of winning away at the Etihad: improbable, statistical chaos, and therefore deeply amusing."

Sutton did not find it quite so funny.

"The game's gone," he replied. "AI will be winning the Premier League soon, at this rate."