City have won seven out of seven and earned fourteen points. The bottom four have earned twenty-two more than they hold

Seven matches, seven wins, 21 points. Manchester City have the kind of start that ends arguments, and on the evidence of the table there is nothing left to discuss.

On the evidence of the chances there is quite a lot.

Replay those seven matches counting only what each side created and conceded, and City finish on 14,5 points instead of 21. The gap of 6,5 is the largest in the Premier League. The leaders are also, by this measure, the most flattered team in the division.

How a perfect record gets built

City have scored 19 goals from 16,6 expected, which is a small bonus and nothing scandalous. The bigger number is at the back: 7 conceded from 9,3 expected, and more importantly a set of results that kept landing on the right side.

They have taken the ball into the opposition box 68 times and allowed 32. Those are the numbers of a very good side. They are not the numbers of a side that wins every single week, and the difference between those two things is worth about two months of league table.

None of this is an accusation. A team that wins seven in a row is doing something right, repeatedly, under pressure. It is a warning about what happens when the coin stops landing the same way, and the first draw arrives looking like a crisis when it is actually a correction.

Arsenal sit on 15 points from 12,0 expected, Chelsea on 13 from 9,4 after scoring 17 goals from 11,9. The top three are all living a little above their means. Liverpool, on 12 from 10,5, are the closest thing to an honest set of books near the top.

The other end, where the bill is enormous

Nottingham Forest have 5 points. The chances they created and conceded are worth 11,4. They have the second meanest defence in the league by expected goals against, 7,3 in 6 matches, behind only Arsenal. They have taken the ball into the box 45 times and allowed 25. Every structural number says upper half, and they are nineteenth.

Bournemouth have 3 points from 9,4. Tottenham have 3 from 7,7, with 3 goals scored in 6 matches from 7,9 expected: no side in the division has lost more in front of goal. Sunderland have 4 from 8,5.

Put those four together and they hold 15 points where the football says 37. Twenty-two points left on the grass by four clubs, which is more than the entire gap between first and seventh.

Fulham deserve a line of their own: 5 points from 9,6, and 74 balls carried into the opposition box, the fourth highest figure in the league. They arrive, repeatedly, and then nothing happens.

What this is worth, and what it is not

Expected goals measure where shots are taken from. They do not know who is standing in the goal, and in a six match sample a goalkeeper in form can move a team four points on his own. When Forest concede 8 goals from 7,3 of chances, the model has very little to complain about; their problem is scoring 5 from 12,2.

So the honest version is narrower than "City are lucky and Forest are unlucky". It is this: the processes at the bottom of this table are not relegation processes, and the process at the top is not a 100% process. Over thirty more matches both tend to drift back towards what they produce.

If City are still winning everything in March with these underlying numbers, the problem belongs to the model and we will write that too. But the direction of travel between now and then is not a guess, it is arithmetic.

Where it bites first

The fixture to watch is whoever meets Forest next. A side with the second best defensive numbers in the league and one win on the board is not a free three points, whatever the table says, and the market will keep pricing them as if it were for a few more weeks.

The same logic runs the other way for City, who at some point will produce a thoroughly ordinary afternoon and draw it, and half the country will call it a wobble. It will be the numbers catching up, nothing more.

T.L.