Gameweek 9 Predictions: Our AI Backs Arsenal at Anfield, and Doesn't Call a Single Draw

Published Friday 30 October 2026. Probabilities pulled 30 October 2026 and may move as team news lands. All kick-off times are UK time.

FootyPulse's AI gives Arsenal a 40% chance of winning at Liverpool on Sunday, against 32% for the hosts and 28% for a draw. It is the model's strongest away-win call of the weekend, and it sits alongside an oddity that only our own data can show. Across all ten Premier League fixtures in Gameweek 9, the model calls eight home wins, two away wins and no draws, yet its own probabilities add up to roughly three draws.

This preview covers every fixture, flags the team news that is driving the numbers, and singles out the calls the model is most and least sure about. Every figure below is FootyPulse's pre-match output for 90 minutes only.

Gameweek 9 predictions at a glance

The standout call: Arsenal at Anfield

The marquee fixture is where the model departs most clearly from the home-win default. Liverpool v Arsenal comes out at 32% / 28% / 40%.

Both clubs carry absence lists into it. Our feed lists seven Liverpool players as unavailable, including Hugo Ekitike, Jeremie Frimpong, Giovanni Leoni and Conor Bradley, with Dominik Szoboszlai doubtful. Arsenal have six listed, but the names matter: Bukayo Saka and William Saliba are missing and Max Dowman is suspended, with Noni Madueke, Riccardo Calafiori and Christos Tzolis doubtful. The model sides with Arsenal anyway — presumably reading their remaining depth more kindly than Liverpool's. Check the Liverpool and Arsenal team hubs for the latest.

The caveat is the margin. Eight percentage points separate Arsenal from Liverpool here, and only six points separate the clubs in the table (Arsenal second, Liverpool fourth). This call also rests on absence data that can change before kick-off.

Biggest predicted upset: Everton at St James' Park

The other away call is Newcastle United v Everton on Monday night: 34% / 28% / 38%. Home advantage would normally make Newcastle the default, but the model points to a soft home record — one win in four at St James' Park this season — and to Everton having won on their last two visits, 3-1 in August and 3-2 last February.

Treat this as a lean, not a conviction. The lead is four points, and our feed currently lists no absentees for either club.

Too close to call: Aston Villa v Fulham

Aston Villa v Fulham is the flattest three-way split of the weekend at 38% / 32% / 30%. Just eight points separate the likeliest and least likely outcomes. The model still leans Villa, but 38% is barely better than one in three.

Villa arrive off back-to-back wins over Manchester City (2-1) and at Newcastle (1-0), but list three players unavailable — André Onana, Aaron Wan-Bissaka and Matteo Ruggeri — while Fulham are without the suspended Ryan Sessegnon, with Emile Smith Rowe doubtful. Two other fixtures run it close, because the call leads the draw by just four points: Chelsea v Manchester United and Hull City v Ipswich Town, both 38% / 34% / 28%.

The most confident call: Tottenham v Crystal Palace

At 58% / 24% / 18%, Tottenham v Crystal Palace is the model's most confident prediction of the gameweek. It rests on form — Spurs won at Chelsea and beat Coventry 3-1 at home in the last two rounds — and it comes with a flag: that Coventry result was Spurs' first home league win of the season in four attempts.

The honest bit: what the model's calls are hiding

• Calls and probabilities disagree. Add up the draw probabilities and the model expects roughly three draws this gameweek (2.95). It calls none. It expects about 4.3 home wins and 2.75 away wins, but calls eight and two. A call is simply the single likeliest outcome in each match, and that rounds every close one towards a winner. If you want the model's real view, read the percentages, not the labels.
• Team news can change. The absence counts come straight from our team-news feed and can change before kick-off. Check confirmed lineups roughly an hour before each game.
• One gameweek proves nothing. We publish our hit rate on the accuracy page precisely because a single preview is not evidence. These are probabilities, not guarantees.

Match-by-match

Chelsea v Manchester United, Sat 12:30. 38% / 34% / 28%. The model leans Chelsea while being sceptical of their form: a 5-1 win over Bournemouth and a 2-0 win at Everton, then a 0-2 home defeat to Tottenham last weekend. United arrive winless in five and without an away league win all season. Both clubs list absentees — Chelsea seven, including Reece James and João Pedro with Moisés Caicedo doubtful; United five, including Benjamin Šeško, Luke Shaw and Manuel Ugarte. A four-point lead over the draw makes this one of the weekend's closest calls.

Hull City v Ipswich Town, Sat 15:00. 38% / 34% / 28%. Hull have eight players listed unavailable, including goalkeeper Jack Butland and defender Matt Targett, with Nampalys Mendy and Loum Gourna-Douath doubtful. They have won one of five at home this season, scoring 1.4 goals per game at the MKM. The model still leans home, but only just.

Coventry City v Sunderland, Sat 15:00. 48% / 29% / 23%. Both sides arrive in poor form, with Coventry losing four of their last five. Sunderland are winless away this season — one point from four trips, seven goals conceded — which tips it.

Manchester City v Brighton, Sat 15:00. 48% / 29% / 23%. City have won all four home games, scoring 11 goals in the process, yet the model gives them only 48%. Brighton list eight players unavailable, including Kaoru Mitoma; City list two, Antoine Semenyo and Matheus Nunes.

Brentford v Nottingham Forest, Sat 15:00. 48% / 29% / 23%. Brentford score 2.00 goals per home game, but five players are listed unavailable, including the suspended Yehor Yarmolyuk.

Bournemouth v Leeds United, Sat 15:00. 48% / 28% / 24%. Bournemouth arrive on a 3-0 away win over Manchester United and a 2-0 home win over Sunderland. The model weights that recent run above their early-season struggles.

Tottenham v Crystal Palace, Sat 17:30. 58% / 24% / 18%. See above: the most confident call, built on form against a modest home record at Tottenham.

Aston Villa v Fulham, Sat 20:00. 38% / 32% / 30%. See above: the flattest split of the gameweek, with Villa in form but carrying absentees.

Liverpool v Arsenal, Sun 16:30. 32% / 28% / 40%. See above: the standout call.

Newcastle United v Everton, Mon 20:00. 34% / 28% / 38%. See above: the biggest predicted upset. Newcastle drew 2-2 with Liverpool at home earlier this season, which the model reads as a credible point rather than a turning point.

Team news watch

Counts include players listed as missing or doubtful in our feed as of Friday afternoon. Coventry, Sunderland, Bournemouth, Leeds, Nottingham Forest, Ipswich and Crystal Palace each list between two and five; neither Newcastle nor Everton has anyone listed.

What's next

Once Newcastle v Everton finishes on Monday night, we will score all ten calls against the results in our Gameweek 9 verdict, misses included. You can follow the running record any time on the accuracy page and see how we compare with human forecasters on AI vs Pundits.

FAQ

Who does FootyPulse's AI predict to win Liverpool v Arsenal? Arsenal. The model gives Arsenal a 40% chance, Liverpool 32% and a draw 28%. Both clubs carry absence lists into the game — seven listed for Liverpool, six for Arsenal including Bukayo Saka and William Saliba — and the eight-point margin is thin enough to move with team news.

Which Gameweek 9 match is closest to call? Aston Villa v Fulham, at 38% Villa, 32% draw and 30% Fulham. That is the flattest three-way split of the weekend.

Which Gameweek 9 prediction is the most confident? Tottenham v Crystal Palace, with Spurs at 58%.

When are the Gameweek 9 matches? From Saturday 31 October (12:30 UK) to Monday 2 November (20:00 UK). Liverpool v Arsenal is on Sunday 1 November at 16:30.

---

FootyPulse predictions are for information and entertainment only. If you bet, you must be 18+. Gambling can be harmful: support is available from GamCare and the GAMSTOP self-exclusion service.