Tottenham vs Newcastle: when our model backs the away team
Published
Most previews start from the home side. This one cannot, because for Newcastle United’s visit to Tottenham Hotspur Stadium on Saturday 29 August our model backs the away team: 36.3% Tottenham, 25.7% draw, 38.0% Newcastle in its 22 August run. The current version of those numbers is on the match page, and it will have moved by kick-off. A 1.7-point edge is not a strong opinion. It is an unusual one, though, because a model carrying a home-advantage term has decided this particular visitor is good enough to overcome the ground anyway. This post is about how large a gap that takes — and about why this instance of it is the shakiest number on our Premier League board.
What home advantage is worth in our numbers
Our model does not publish a “home bonus” as a figure, but you can bracket it from two fixtures in the same run.
Take today’s opener at St James’ Park. Our team ratings, updated 22 August, have Liverpool on 1658 Elo and Newcastle on 1564 — the visitors 94 points the better side. The model’s split for Newcastle vs Liverpool is 38.0% home, 25.0% draw, 37.0% away. Ninety-four points of away superiority, and the home team still comes out a point ahead. So the venue is worth more than 94 rating points.
Now this fixture. Newcastle are on 1564, Tottenham on 1463 — the visitors 101 points the better side, one week later, same model. The split is 36.3/25.7/38.0, and this time the away team edges it by 1.7. So the venue is worth less than 101 rating points.
That brackets our model’s home-field effect at roughly 95 to 100 rating points in the Premier League. It is an approximation rather than a constant — the model prices attack and defence separately, so Elo alone does not determine a fixture — but as a rule of thumb it holds: a visiting side needs to be about a hundred points better to be favoured on the road, and Newcastle clear that bar by one point.
Which is exactly why the call is worth interrogating rather than repeating. It survives on a margin thinner than the precision of the inputs.
The rating Newcastle are beating
Tottenham at 1463 is the lowest rating of any club named in this post, and the model did not arrive there capriciously. Spurs escaped relegation last season after a second consecutive 17th-place finish, as Sports Mole’s season preview sets out, having got through three head coaches on the way. Roberto De Zerbi arrived on 31 March 2026 as the third of them, on a five-year deal, per Sky Sports, with the club a point above the bottom three and seven games to play.
A ratings model reads that season as what it was: 38 matches of a bottom-four-adjacent team. The rating is not an insult, it is arithmetic. Our attack and defence figures say the same thing more precisely — Tottenham at −0.002 for attack and −0.085 for defence, a side that scores at roughly league average and concedes above it.
The squad that earned that rating has been rebuilt
Here is the problem with using it. Tottenham signed Sandro Tonali from Newcastle for a club record — £92.5m plus £7.5m in add-ons, reported by Sky Sports on 2 July 2026 — and the squad now listed on Tottenham’s FotMob page carries Andrew Robertson, Kevin Danso, Marcos Senesi, Jan Paul van Hecke and Martin Dúbravka among the defensive and goalkeeping options.
Our model does not read transfer news. Not “weights it lightly” — does not read it. The 1463 was computed from results, and the results were produced by a different group of players under two of De Zerbi’s predecessors. It will only update once this team plays matches, and by 29 August it will have exactly two of them: the opener away at Brentford on 22 August, and this.
That is the honest shape of the uncertainty here. The model is not wrong to rate Tottenham low; it is simply answering a question about last season while the fixture asks one about this season.
Newcastle’s own subtractions
Symmetry matters, and Newcastle have lost more than they have gained on the specific evidence available. They sold the £100m midfielder to the side they are about to visit, and they arrive with a new manager in Andoni Iraola. The Shields Gazette’s injury round-up before the Liverpool opener listed Tino Livramento out with a hamstring problem picked up at the World Cup, Joelinton unavailable, Lewis Miley only just back from a broken leg, and Fabian Schär recovering from a groin injury.
Four absences point one way and a £100m sale points the other, and our 1564 has priced none of it. The reason for saying so plainly is that a reader might otherwise assume a number this specific must have accounted for them. It has not.
What we would and would not do with 38.0%
We would not call this a pick. A 1.7-point separation between the two sides is the model saying it cannot split them, and the 25.7% draw is doing more work than either favourite label suggests — the single likeliest scoreline is 1-1 at 11.8%.
The clearer statement on this fixture is the goals line, not the result: 59.1% for over 2.5 goals and 62.3% for both teams to score, driven by two defences our ratings put below average. If you want the model’s view of this match in one sentence, it is “no idea who wins, reasonably sure it is not a 1-0”.
Whatever happens, the call gets graded either way on the public track record, the rest of the round sits on the Premier League hub, and the ratings behind all of this update on the Tottenham and Newcastle team pages as the season fills them in.