Gameweek 8 Predictions: Our AI Backs Arsenal and Chelsea at 68% — and Sunderland v Leeds Is a Dead Heat

FootyPulse's model puts Arsenal v Everton and Chelsea v Tottenham joint-top of Gameweek 8, with a 68% chance of a home win in each. At the other end of the scale, Sunderland v Leeds is so tight that home win (34%), draw (32%) and away win (34%) are separated by just two points.

Those are the two bookends of a weekend that runs from Ipswich v Nottingham Forest on Friday night to Sunderland v Leeds on Sunday afternoon. Here is every fixture, with the AI's win/draw/away split, the calls worth your attention and the uncertainty behind each call.

All probabilities are for 90 minutes and are as of Thursday 22 October. They refresh daily, so check each match page before kick-off.

The Gameweek 8 numbers at a glance

Add up all 10 fixtures and the model's average split is 44.6% home win, 26.4% draw, 29.0% away win. Read as expected counts, that is roughly 4.5 home wins, 2.6 draws and 2.9 away wins across the round. That is simple arithmetic on the percentages, not a forecast of the exact outcome.

Six fixtures lean home (Ipswich, Arsenal, Chelsea, Liverpool, Manchester United and, narrowly, Coventry). Three lean away (Manchester City, Newcastle, Brentford). One, Sunderland v Leeds, has no clear lean at all.

Four of the ten fixtures have a gap of five points or fewer between the two likeliest outcomes. This is a weekend with a lot of genuine uncertainty in it.

*UK clocks go back an hour at 02:00 on Sunday 25 October, so Sunday kick-offs are in GMT.

Standout call: Chelsea v Tottenham and Arsenal v Everton, a dead heat at 68%

There is no clear single standout this week, because two home sides share the top figure.

Chelsea v Tottenham Hotspur edges it on margin: the lead over the next likeliest outcome is 49 points (68% against a 19% draw), one point wider than Arsenal's. Tottenham's chance of winning the derby is 13%.

Arsenal v Everton is the most lopsided on the away side: Everton's 12% is the lowest visiting figure of the weekend.

One point separates them. We would not build a case on it.

Biggest predicted upset: there isn't a true one, but here is the closest thing

The probabilities do not establish which side is an underdog against an independent benchmark. Here, "upset" simply means an away lean, not a claim about market expectations.

If you define it as the model siding against the home team, the biggest swing is Manchester City at Aston Villa, where City are given 58% and Villa 18%. City are favourites on reputation as well, so that is a road call rather than a shock.

The more contrarian leans are smaller. The model gives Newcastle at Crystal Palace a 38% edge over Palace's 28%, a ten-point gap. At Hull v Brentford the visitors lead by just four (38% to 34%), so a home win would not be a surprise.

Too close to call: Sunderland v Leeds

Sunderland v Leeds United is the tightest fixture on the board: 34% Sunderland, 32% draw, 34% Leeds. Home and away are level, and the draw sits only two points behind both.

It is one of only two fixtures this weekend where the draw reaches 32% or more. Crystal Palace v Newcastle has the highest draw figure of the round at 34%.

Team news before kick-off

Player absences and historical form claims in the supplied analysis could not be independently verified for this round, so they are not presented here as confirmed news. Check club updates and the Premier League injury list, then return to the live Gameweek 8 fixtures for the latest analysis.

Match-by-match: Gameweek 8

Ipswich Town v Nottingham Forest — Friday 23 October, 20:00

52% / 26% / 22%. A clear but not overwhelming lean to the hosts. Ipswich lead the draw by 26 percentage points, but the combined draw and away probability is still 48%. Full prediction →

Aston Villa v Manchester City — Saturday 24 October, 12:30

18% / 24% / 58%. The biggest away lean of the round, and the lowest home figure. City lead the next-likeliest outcome, the draw, by 34 percentage points. Full prediction →

Arsenal v Everton — Saturday 24 October, 15:00

68% / 20% / 12%. Joint-highest home figure. The home win sits 48 points above the draw. Everton's 12% is the lowest away chance on the card. Full prediction →

Coventry City v Fulham — Saturday 24 October, 15:00

38% / 29% / 33%. Narrow home lean, with only five points between Coventry and Fulham. The draw is only nine points below the home win, so neither side has a commanding probability. Treat this as close to even. Full prediction →

Chelsea v Tottenham Hotspur — Saturday 24 October, 17:30

68% / 19% / 13%. The widest margin of the weekend. The draw and away win together account for 32%, so a home win is far from certain. Full prediction →

Crystal Palace v Newcastle United — Sunday 25 October, 14:00

28% / 34% / 38%. The model leans narrowly to Newcastle, but the draw at 34% is the highest of the weekend. Only four points separate the away win and the draw. This one could go any of three ways. Full prediction →

Hull City v Brentford — Sunday 25 October, 14:00

34% / 28% / 38%. The home and away probabilities are close, with the draw six points below Hull. Four points separate the sides, so this is a lean, not a call. Full prediction →

Liverpool v Brighton & Hove Albion — Sunday 25 October, 14:00

48% / 28% / 24%. Liverpool are favourites, but 48% is the lowest figure among the clear home favourites, and the combined draw and away probability is 52%. Full prediction →

Manchester United v Bournemouth — Sunday 25 October, 14:00

58% / 24% / 18%. A solid home lean: United lead the draw by 34 points, but there is still a 42% chance of a different outcome. Full prediction →

Sunderland v Leeds United — Sunday 25 October, 16:30

34% / 32% / 34%. A genuine three-way split, with only two points between the most and least likely outcomes. If you want one fixture where the AI is shrugging, this is it. Full prediction →

How to read these numbers

A well-calibrated 68% estimate implies a different outcome roughly one time in three over many comparable matches. Both of our joint-standout fixtures can and will occasionally go wrong, and the same goes for every other fixture on this page. These are probabilities for 90 minutes, not promises.

What we can promise is that we publish our record. You can see how the model has actually performed on our accuracy page, including the misses.

Predictions also move as team news lands, so the figures above are a Thursday snapshot. After the final whistle on Sunday we will publish the Gameweek 8 verdict, with the hits, the misses and how the AI compared with the BBC's pundits on our AI vs Pundits page.

FAQ

Which Gameweek 8 match is the closest to call?

Sunderland v Leeds United. FootyPulse gives Sunderland 34%, the draw 32% and Leeds 34%.

Which away team does the AI favour most in Gameweek 8?

Manchester City at Aston Villa, at 58%. Newcastle at Crystal Palace (38%) and Brentford at Hull City (38%) are narrower away leans.

What time do Gameweek 8 matches kick off?

The round runs from Ipswich v Nottingham Forest on Friday 23 October at 20:00 to Sunderland v Leeds on Sunday 25 October at 16:30. UK clocks go back on Sunday morning, so Sunday kickoffs are in GMT. Friday has one match, Saturday four and Sunday five.

Are FootyPulse predictions guaranteed?

No. They are probabilities for 90 minutes. A well-calibrated 68% estimate implies a different outcome roughly one time in three over many comparable matches. See the full scored record, including the misses.