“They had three big chances and scored two goals, we had lots of chances and only scored one. We got into a lot of good positions, a lot of coaches might say that we should have, could have won, but in the end you have to do it and we didn’t do it. We have to look at a couple of other little issues to make sure that when we do get into good positions, that we do score goals.”
This was [part of] Joe Montemurro’s assessment after the last-gasp defeat to Chelsea in the Conti Cup Final. Arsenal dominated territory and created more chances, but Chelsea were more clinical and won. It was a similar story three weeks earlier when the Gunners went to Manchester and dominated City but were left to rue missed chances as they fell to a damaging 2-1 defeat.
I asked Jordan Nobbs if chance conversion was becoming an issue a few minutes after I posed the question to Joe. “We can maybe get away with it in some games, but when we play teams like Chelsea, we have to be on it and take our chances,” she admitted. Tellingly, she critiqued her own performance in this respect, “I think maybe I need to shoot a little bit more at times.”
Five days earlier, I spoke with Joe and Jordan ahead of the Conti Cup Final. I had been perusing Statsbomb’s mine of WSL data to see if my suspicion of Arsenal’s profligacy held water. The data certainly reveals that Arsenal shoot less often than Manchester City and Chelsea but take marginally better-quality shots when they do chance their arm.
|Shots per game||Shots on target per game||Differential|
This is a consequence of Montemurro’s precision, possession-based style. Arsenal favour the scalpel approach and are not afraid to go backwards to go forwards. I asked Montemurro about this at the pre-Conti Cup Final media day. “I hate crosses for crossing’s sake,” he explained. “If it’s not on to cross and there’s no one in the box, then we start again and we try to probe and find the space.”
It isn’t only games against Chelsea and Manchester City that have given the impression of profligacy. Home victories against Birmingham [2-0] and Liverpool [1-0] and the away victory at Spurs [2-0] ought to have been won more convincingly as Arsenal pushed their food around the plate for large sections of the game. Looking at the data- which only takes in WSL games- Arsenal’s XGD [expected goal difference] is nestled somewhere between Manchester City and Chelsea’s.
|XGF [Expected goals for per game]||Goals per game||XGC [Expected goals against per game]||Goals against per game||XGD [Expected goal difference]||Goal difference per game|
|Arsenal||1.90||2.66 [+0.76]||0.68||0.86 [+0.18]||1.21||1.80 [+0.59]|
|Chelsea||2.49||3.13 [+0.64]||0.56||0.73 [+0.17]||1.92||2.40 [+0.48]|
|Manchester City||2.02||2.43 [+0.41]||0.87||0.56 [-0.31]||1.16||1.87 [+0.71]|
Yet the data doesn’t suggest Arsenal are any more profligate than their title rivals. In fact, they are overperforming their XG slightly more than City and Chelsea. I wondered how much the figures had been skewed by the Miedema inspired 11-1 win over Bristol City- a day in which the Gunners scored 11 from an XG of 5.25. I wanted to know how many times Montemurro’s side had underperformed on XG compared to City and Chelsea.
I was surprised by the results. Arsenal have underperformed their XG in seven of their 15 WSL games, three of those occasions by more than 0.5 goals. Chelsea have underperformed on XG six times in 14 games and in five of those games it was by more than 0.5 goals. City are the same as Arsenal, scoring below their XG tally on seven occasions, but only twice by more than 0.5 goals.
However, the Gunners scored once from an XG of 1.51 in a 2-1 defeat at City last month [the home side’s two goals came from an XG of 0.63], they scored once from an XG of 1.44 in the 1-0 victory over the same opponents in October. Unfortunately, Opta do not produce stats for the Conti Cup, but I am willing to bet that Arsenal’s XG was more than 1.5 in the 2-1 reverse to Emma Hayes’ side in the final.
Maybe the fact that Arsenal’s shot volume is lower means misses are more keenly felt, compared to Chelsea’s more blunderbuss approach to attack. I wanted to examine whether individual players are skewing the team stats, in Vivianne Miedema Arsenal have arguably the finest centre-forward in the world. I looked at the XG of Arsenal’s principal attackers. To what extent is Viv boosting the stats?
|Shots per 90||XG per 90||Non-penalty goals per 90|
|Danielle van de Donk||1.37||0.27||0.33 [+0.06]|
|Jordan Nobbs||2.46||0.28||0.41 [+0.13]|
|Kim Little||1.11||0.14||0.32 [+0.18]|
|Vivianne Miedema||3.90||0.67||1.09 [+0.42]|
|Beth Mead||2.59||0.26||0.24 [-0.02]|
Well, she is a little, but Danielle van de Donk, Jordan Nobbs and Kim Little are all overperforming their XG to varying degrees, with Beth Mead very slightly underperforming hers. Unfortunately, this level of data does not exist from Arsenal’s league winning campaign in 2018-19, so it’s not possible to compare. The impression remains that the Gunners have been wasteful in front of goal, but the numbers don’t really bare it out.
If the data existed for cup games, I tend to think a slightly different picture might emerge. I am certain Arsenal underperformed XG in Conti Cup knockout ties against Reading [1-0] and probably even Manchester City [2-1] and the final against Chelsea. The group stages also featured their solitary scoreless game this season against Brighton. The data we do have confirms that Arsenal have a more precise style resulting in fewer shots, but better-quality shots than their immediate rivals. But the numbers do not suggest that chance conversion is a big issue.
|Fixture||Total Shots||Shots on target||XG [expected goals]||Goals scored|
|West Ham United [H]||17||5||2.06||2|
|Manchester United [a]||21||9||1.61||1|
|Brighton & Hove Albion [H]||12||5||1.97||4|
|Manchester City [H]||10||3||1.44||1|
|Tottenham Hotspur [a]||19||3||1.80||2|
|Bristol City [H]||35||16||5.25||11|
|Everton [a]||15||8||3.20 [including penalty]||3|
|Birmingham City [H]||12||7||2.39||2|
|Brighton & Hove Albion [a]||14||5||2.36||4|
|Manchester City [a]||15||7||1.51||1|
all data derived from Statsbomb