Mikel Arteta seems to have restored a feel-good factor among Arsenal fans. His no-nonsense approach and clarity of communication contrast starkly with the seemingly confused state of affairs – both on the pitch in terms of performances and off it in press conferences – that existed under Unai Emery.
There’s also a sense that performances have improved, but the most telling factor is obviously results. Nothing engenders a positive outlook more than a few wins. The problem is that results can be deceptive and underlying performances could be telling us a different story.
After similar upticks in form in recent seasons for the Gunners, things have gone awry. A deeper look into expected goals should help determine whether things really have been changing under Arteta.
By measuring chance quality, xG can show how many goals a team should have scored or conceded compared to the actual numbers and give us a sense of whether a team is performing to expectations, over performing, under performing or whether luck plays a factor.
Back in 2016/17, Arsenal were struggling in defence. After 30 Premier League games, they were conceding goals at a rate of 1.3 per match. xG told us they should be conceding around 1.2 per game, so things were about as bad as they seemed.
Arsene Wenger then decided to switch to a back three and things changed. The Gunners started to concede at a rate of just 0.7 goals per game in the final eight matches of the league season, plus the two cup ties against Man City in the semi-final and Chelsea in the final. Wenger’s change in tactics was hailed as a masterstroke.
Expected goals hadn’t received much mainstream media use at that time, but it was clear when we looked at the data that things were not as good as they seemed with the Gunners’ defence. They were in fact worse in the context of xG with the team conceding 1.6 expected goals per game – more than twice what the team were actually conceding and even worse than they had been doing with a back four. What had happened over a few games was not that the defence was better, it was simply that Petr Cech had an impressive stretch in goal, the opposition strikers lost their shooting boots and Arsenal essentially rode their luck. As the 2017/18 season got underway, it soon became clear little had changed as the Gunners conceded at a rate of 1.34 per game during the campaign, while xG was 1.27 and Wenger lost his job after his side conceded 51 goals, a club record in the Premier League era.
In 2018/19, Emery’s Arsenal went on a 22-match unbeaten run. xG showed Arsenal were scoring at an unprecedented rate, netting 50 goals, but with xG at around 33. This level was clearly unsustainable. The ball stopped finding the net and shots were saved or found their way into the crowd. Arsenal’s attack suddenly stopped bailing out the defence and the top four and European glory slipped away.
And so to Arteta’s opening 15 games since his December appointment. Arsenal have conceded 0.8 goals per game, compared to 1.45 per match under Emery, which sounds good. When you look at shots allowed, Emery’s Arsenal allowed 15.8 shots per game and Arteta’s Gunners have given up only 12.5 per game. Again, seems like a definite improvement.
However, when you look deeper, Emery’s team conceded 29 with an xG against value of around 28 – in other words as you’d expect. But when you look at Arteta’s 15 games in charge, the Gunners have shipped just 12 goals, with an xG against of around 20. Some poor finishing by opponents – Burnley’s Jay Rodriguez, Calvert Lewin for Everton and Michail Antonio for West Ham – and the performances of goalkeeper Bernd Leno have turned defeats into draws and draws into wins.
It’s a similar story on quality of chances. The average xG value of chances given away under Emery is 9% (0.09 xG) – so roughly one in 11 might expect to be scored. It’s 11% under Arteta – one in nine. This means teams should actually require fewer chances to score against Arteta’s team than Emery’s team because they have been on average of a higher quality.
Arsenal’s opponents have had a higher xG in seven of the 15 games under Arteta, and that includes three of the last four Premier League games against Burnley, Everton and West Ham.
The bad news is Arsenal will start to concede lots more goals if things carry on the way they have gone so far.
The good news is that expected goals can help predict what will happen if a team keeps doing the same things. Arsenal should know that they have been a bit fortunate at times. They will know that their defence hasn’t miraculously improved overnight with no change of personnel.
The new regime will know they will have to change their approach in defence to address the underlying numbers and not settle for headline goals conceded figures. A long-term fix is still needed.