The 2026/27 Premier League Table Without 'VAR Errors': When Fan Votes Wear the Clothes of Data

On August 13, 2026, a different Premier League table scrolled across my phone. Not the league's own. It was the table Squawka rebuilt after subtracting every VAR error from the opening three matchdays. At the bottom, Tottenham Hotspur sat twentieth with zero points.

I read it three times. Tottenham spent over £300m in a single transfer window, played three matches and scored no goals. Nottingham Forest climbed five places on the back of one disallowed goal. Manchester United sit in the lower half of the real table yet fall inside the corrected top ten — without gaining a single point. Arsenal and Manchester City did not move at all: three games, three wins, no disputed decisions.

What stopped me was not the table. It was the question behind it: what data produced this table, and how much weight can that data carry?

Context: a format borrowing the authority of a league table

English football's VAR argument has run since 2026/20 and reached an acceleration phase in 2026/27. Roy Keane and Wayne Rooney publicly called for abolition. A major football media brand built a recurring product around it: each matchday, supporters vote on which decisions were wrong, and the table is redrawn.

Two layers need separating. The first is result data — who won, who lost, by how many goals. The second is the ballot — who believes the referee erred. The 'no VAR errors' table blends both and presents the output in the visual grammar of a standings page. That format borrows the credibility of an official document: rankings, points, an ordering from top to bottom. The contents inside are not that.

Across four years working in player development and tracking youth data, I learned one rule: when a dataset does not publish its sample size, the reader must ask what tests it can survive. This table publishes no vote count, no geographic distribution, no neutral weighting. That is the first blind spot.

Core: the hardest signal available is a negative one

The hardest signal in the entire dataset is a negative one: a club that spent more than £300m in a single window and then failed to score across three openers is showing a conversion failure at the end of its build-up chain. But three matches is a statistically negligible sample. It can be a start; it cannot be a trend.

A larger problem: no xG, xA or PPDA figure exists anywhere in this dataset. There is no way to distinguish Tottenham failing to create from Tottenham creating and failing to finish. Those two conditions demand opposite remedies — one is a structural creativity problem, the other a finishing and variance problem. The honest analytical position is that the causal layer is empty.

Roberto De Zerbi's coaching profile, cited as Tottenham manager, implies a possession-heavy, control-oriented build-up. That family of systems typically carries a high integration cost early in a project cycle, because it demands positional discipline and rehearsed patterns. Three scoreless matches are consistent with a system still being installed — but they do not prove it.

Leeds United under Daniel Farke belong to a different category entirely. Leeds appear here as a beneficiary of the real table and a loser of the corrected one, after a penalty not awarded against Brentford. That is the cost of a decision, not evidence about Leeds' playing model.

And here the dataset betrays itself. The entire bottom-of-table reconstruction rests on a single call: the disallowed Neco Williams goal. Forest gain two points and five places; Tottenham lose ground. That is a single-point-of-failure dataset — and at matchday three, with the whole division compressed into a few points, any single correction reshuffles multiple positions at once.

Contrarian angle: the corrected table is not a correction

The contrarian angle sits here. The 'corrected' table is not a correction. It is a second opinion, weighted by supporter sentiment.

Separate two kinds of VAR error. The first is factual: a miscalibrated offside line, a wrongly identified player. That is checkable with data. The second is a threshold dispute: should VAR have intervened at all, under the 'clear and obvious' bar? That is a judgement call, and fan voting is structurally unsuited to adjudicating it.

The Williams incident belongs to the second kind. The Premier League issued a same-day statement citing handball. The objectors said the footage was inconclusive and not clear and obvious. The two sides are not arguing on the same rule layer — one discusses factual determination, the other the intervention threshold. A table built from ballots may have merged both.

One more sampling mechanism: only disputed incidents go to a vote. Correct decisions nobody contests are excluded from the sample by construction. That selection bias systematically inflates the apparent level of controversy. Supporters of the disadvantaged club have a higher motivation to vote than neutrals, and nothing in the methodology corrects for it.

Meanwhile, the firmest section of the piece sits at the top. Arsenal and Manchester City won all three matches and were party to no disputed decision. That is the article's own internal control group: there, the real table and the corrected table coincide. Where a method has nothing to correct, results do not diverge. That is the most trustworthy detail in the entire dataset.

As for Manchester United: they enter the corrected top ten without gaining a point. The position improves purely because others drop. A table can lift a club's ranking without lifting its results — a small detail that says a great deal about the format.

I still remember the lesson from the 2026 World Cup. After the group stage I wrote that Mbappé would win it. In the quarter-final on July 6, Uruguay kept an average of 7.8 players behind the ball and closed every space behind the defensive line; Mbappé completed no successful dribble in the first thirty minutes. I rewrote the piece. Uruguayans do not build walls. They build a manifesto about space. This corrected table does the same thing: it does not build a new standings page, it builds a manifesto about justice — easy to read, easy to share, and easiest of all to misread.

Takeaway: the first brick must be a number, not a ballot

Underneath the raw data, I found the first brick of a generation — but that brick has to be a verifiable number, not a vote. Home advantage was once a fortress, and the pandemic taught me the fortress is only a variable; a league table is the same. It is not truth, it is a variable dependent on the method that built it.

If the no-VAR-errors table makes you see Tottenham differently, hold that question for three months. If they are still scoreless at matchday ten, that is data. For now, the only thing we know for certain is that Arsenal and Manchester City are winning — and they are winning without anyone redrawing the table on their behalf.