Arsenal head to Nottingham Forest on Sunday sitting second in the Premier League, three points behind Manchester City.

Enzo Maresca’s side have won all six of their league games so far and host Ipswich on Saturday, while Arsenal’s only slip came in the 3-0 defeat at Brighton before the international break.

On the pitch, they’ve been setting the pace, but when you look a little closer at the underlying data, Arsenal have no reason to panic yet.

Underlying numbers suggest Arsenal are by far the best team in the Premier League

According to Opta data shared by Nick Wright of Sky Sports, Arsenal have the best expected goal difference (xGD) per game in the league by some distance. The Gunners boast a figure of 1.02, with City next on 0.62 and Brighton third on 0.59.

For context, Arsenal won the title last season with an expected goal difference of 0.97 per game. So despite the defeat on the south coast, Mikel Arteta’s side are performing at a higher level in this (admittedly small) six-game sample than they did during their title-winning campaign. Underlying numbers like these are generally seen as a better predictor of where a team will finish than the league table this early on.

As for City, no team has outperformed their xG more, with Opta Analyst’s expected points model suggesting they should have around seven fewer points than they do. That’s down to their finishing at one end, but also how much they’re giving up at the other. Sunderland created over 3.0 xG from 17 shots against them, and even Coventry managed 12 shots in their 1-0 defeat at the Etihad. Arsenal beat the same Coventry side 3-0 on the opening day and allowed them just four shots.

Of course, there’s an argument that City don’t need as many chances because of the quality they have in front of goal. Erling Haaland was seemingly bred in a lab to score goals, and that’s an edge in attack that Arsenal lack. But if City keep giving up chances like that, their results are unlikely to keep matching their numbers, and Maresca-led sides have a history of fading after fast starts.