A Table And A Model Pulling Apart
The most interesting thing about Tottenham versus Everton is not the venue or the kickoff time. It is that the current Premier League table and Nutmegly's AI read point in opposite directions. Tottenham sit 18th with one point from three matches and zero goals scored; Everton sit eighth with five points and no defeats. Yet Nutmegly's model, generated on 8 September, leans home: it gives Tottenham a 44.1% win probability, the draw 28.3% and Everton 27.6%.
That is not a majority for Tottenham, and it is not a guarantee. It is a leading outcome in a three-way distribution, with more than half the probability mass sitting on a draw or an Everton win. The value of the preview is in asking why the model reaches that conclusion, and whether the current-season evidence is strong enough to overturn it.
The Current Table Points Everton's Way
Start with what has actually happened in 2026. Tottenham have played three, drawn one and lost two, scoring none and conceding five. At home they have played once and lost 0-2 to Newcastle. Away from home they drew 0-0 at Nottingham Forest and lost 0-3 at Brentford. Everton have played three, won one and drawn two, scoring five and conceding three. They drew 2-2 with Manchester United, drew 1-1 at Bournemouth and beat Crystal Palace 2-0. Those are current-season results, not historical padding, and they frame the match as a meeting between a side that cannot score and a side that has not lost.
The broader tracked samples add nuance. Across the last eight recorded matches for each side, Tottenham have taken 1.38 points per match, while Everton have taken 0.88. That is the clearest evidence in the home side's favour, and it helps explain why a model might lean Tottenham even when the current table looks ugly. But those eight-match samples stretch across seasons, as the data notes warn. They are not a three-game form guide, and they should not be read as proof that Tottenham's current attack is functioning.
Why The Model Still Sees A Home Edge
The home case has other supports. Tottenham's most recent recorded meeting with Everton was a 1-0 home win on 24 May, though that belongs to the previous sample rather than this season. The rest window also tilts slightly home: the gap since each side's last recorded match is seven days for Tottenham and six for Everton, a small difference that does not establish a meaningful advantage. The model's home lean, then, is not built on current scoring form. It appears to rest on a mixture of home advantage, the broader points-per-match signal and the fact that a 44.1% home win is the single largest outcome in a volatile distribution.
The goal projection is similarly cautious. Nutmegly's model assigns less than 40% to over 2.5 goals, at 39.8%, so the lower-scoring side of that line carries the larger probability. That fits a fixture where the home side has yet to score in the league and the away side has drawn two of three. If the match does turn open, it will be because something in the lineups or early game state changes the expected pattern, not because the pre-match numbers point that way.
The Counterevidence Is Serious
The counterargument is straightforward: the current-season evidence is not just slightly against Tottenham, it is heavily against them. Tottenham are 18th with one point; Everton are eighth with five. In the Premier League sample, Tottenham average 0.80 points per match while Everton average 1.00. In the tracked home-away split, Tottenham average 0.80 goals per home match, while Everton concede 1.60 goals per away match — a combination that, on the surface, sounds more like an Everton opportunity than a Tottenham one, albeit from small samples. Everton are unbeaten and Tottenham have not scored.
There is also a model-level caution. The supplied counterpoints note that Nutmegly's internal model and an external forecast disagree on direction, and that the difference may come down to sample definitions. Lineups are not confirmed, and the injury feed available reports no absences. That last point is not a clean bill of health; it simply means no missing players were reported through that channel. With starting lineups unconfirmed, the pre-match probabilities carry more uncertainty than usual.
The Watchpoint: Can Tottenham Score At All?
The match should be judged less by the win probability than by one question: can Tottenham score in the current Premier League season? They have zero goals from three matches. Everton have kept one clean sheet in the current season, in the 2-0 win over Crystal Palace, and conceded twice to Manchester United and once at Bournemouth. If Tottenham's blank continues, the home lean will look like an over-weighting of older samples and home advantage. If they score early, the broader-form argument gains credibility.
The conditional paths are clear enough. A low-scoring game favours the draw or a narrow Everton result on current evidence; a Tottenham goal changes the pressure on the model's home lean. The specific watchpoint is whether Tottenham can turn their first current-season goal into a lead before Everton's unbeaten start has a chance to settle the match. That is not a prediction, just the hinge on which the evidence swings.