Match Review: Fulham 0-Arsenal 3

This review covers Fulham’s 3-0 home defeat against Arsenal. It will use data collected from Infogol, WhoScored and some primary data I collected using the latest available technology (pen and paper) while watching the game!

I like to go into a lot of detail with these reviews, and this may not be to everyone’s taste: but let me start with a summary of what I have found by looking into the numbers and if that whet’s you appetite for more, please do continue!

Summary

The key observations from reviewing this game are:

• The final score is slightly flattering to Arsenal based on balance of play (I think 2-0 would be the most likely outcome based on chances created) but the most deserving team won the game
• Most of Arsenal’s expected goals (about 2/3rds) can be attributed directly to two poor and largely unforced defensive errors by Fulham. Without those errors and all else being equal, a 0-0 scoreline would have been fair. Fulham were therefore fairly resilient in defence (errors aside)
• Arsenal overloaded the left flank and channelled most attacks unsuccessfully through that area.
• Fulham shifted right to compensate, leaving space on the opposite flank (Fulham’s notoriously weak left) which was the source of most of Arsenal’s actual attacking threat. At times Fulham put both wingers (Cav and Kebano) on the right to combat the Arsenal overload
• Fulham put a lot of balls including 16 crosses into the Arsenal box, but were unable to turn many of these moments into clear goalscoring chances
• Indeed Fulham’s attacking xG of 0.16 is the lowest they have ever recorded for a league game!
• Despite criticism on Match of the Day, Fulham’s press worked quite well, limiting the quality of Arsenal’s position and generating a few chances with possession won in the attacking half.
• Anguissa made a big positive impact when he came on, high rate of involvement on the ball, active in attacking areas, tied first in the match for tackles and dribbles (admittedly with only 2 of each) all despite being on the pitch for just 27 minutes.
• Mitrovic struggled to support the pressing game, Parker used Decordova-Reid as a strike partner to help him, but had to release the double pivot to do it.
• Set piece defending for the second goal was poor with two defenders left marking four Arsenal players despite almost no attacking movement on the corner.

Read on to find out more!

Did the outcome reflect the balance of play?

This is something that I will always try to answer when looking at game results and I will usually try to answer this by looking at the expected goals (or xG) for the match.

I have produced the cumulative xG chart below for the match using Infogol data, and I have marked the 3 actual goals with the little footballs.

The red line shows how Arsenal accumulated xG through the 90 minutes. It is punctuated by two very big chances at the start of each half, both of which resulted in goals.

The extremely flat white line is not the x-axis but is in fact Fulham’s cumulative xG, although, admittedly, it does serve as a functional horizontal axis as well.

Let’s not hide away from this, the white line reflects Fulham’s cumulative xG of 0.16, the lowest xG I can find recorded for a Fulham league match ever (I looked back to 2014 which is around when this started getting recorded).

There are a few interesting notes on the xG though,

1. Fulham were given no xG for Karama’s early press and 1 on 1 with Leno. That’s because he didn’t actually get a shot away, but this is clearly a weakness in the xG model because that was obviously a reasonable goal scoring chance, and I would say one could think about adding anywhere up to 0.2xG for that (which would double Fulham’s xG and give us the highest of the 3 promoted teams in gameweek 1 – clutching at straws I know)
2. In the horrendous incident that led to the first Arsenal goal, the Gunners actually took 3 shots, Xhaka had one from the edge of the box (0.04 xG), then in the ensuing scramble Willian had one saved by Rodak (0.68 xG) and finally Lacazette put away the rebound (0.67xG). So Arsenal players accumulated 1.39 xG in a single incident but it is clearly illogical to expect a team to score more than one goal in a single attack! Infogol deals with this by capping the team xG at 1 for the incident (while giving the individual players the full allocation) but this approach undoubtedly overstates the probability of a goal being scored.

So Arsenal’s recorded team xG of 2.19 is probably slightly overstated and Fulham’s 0.16 is maybe a little harsh, but overall, a result of between 2 and 3 to nil is about right. 2-0 is probably the most likely result based on the balance of chances

One final point here though, the first two goals scored by Arsenal featured huge defensive errors (more on this later), I would contend that they did not rely on Premier League level quality attack and could just have easily have been scored by a lower league team. These were largely unforced errors and as error prone as Fulham sometimes are, I think this is an unusual error rate even for them. If these errors had not been made, the Arsenal xG would be about 0.7 (I quoted 0.4 in a highly liked tweet the other day, but I have now updated to reflect my understanding of Infogols treatment of the first goal).

In other words, without the big unforced errors, and all else being equal a 0-0 draw would not have been an unreasonable outcome.

A lot of caveats there, and of course Arsenal would have pushed harder for goals if they didn’t already have them, but that might have created openings for Fulham too.

Tactics: Wonky Wonky Arsenal

Arsenal set up for the game in a 3-4-3 formation, but the system they deployed was strangely asymmetric in possession. As in the playoff final match review I will start by showing the heat maps for the opposing starting XI, set out in formation used:

Starting with the defence we can see that Gabriel and Holding play, in possession at least, as standard Left and Right Centre Backs, despite being part of a back 3 when out of possession.

Kieran Tierney then moves into a left full back position, something that Rob Holding does not attempt to do on the right.

Looking at the midfield we see Maitland-Niles also playing a left-wing position, which means that he and Tierney are effectively trying to double up on the left flank. We can see a different approach on the right where Hector Bellerin does this role alone.

Plan A then, for the Arsenal attack, appears to be to have Tierney and Maitland-Niles force their way through the left flank and feed Aubameyang who tries to come in to attack the box off the left wing. Its a bit like the Sheffield United overlapping centre backs, but only on the left!

When this fails, they try Plan B which seems to be to work the ball across the back and over to Bellerin so he and Willian can get in a position to cross the ball, back over for Aubameyang again!

So plan A and B are not simply attack left and attack right, plan B usually follows an unsuccessful attack on the left, having shifted the opponent to that flank.

Plan A and B are described in the diagram below:

This basic attacking structure means that Arsenal are continually overloaded on the left. Fulham naturally shifted over to respond which made the left side of the pitch (from Arsenal’s point of view) very densely packed. This means that when plan B is used, Bellerin and Willian often find themselves with a lot of space (to try and cross for Aubameyang).

The majority of attacks follow strictly Plan A – the ball goes to Tierney and he and Maitland-Niles try to work the ball to Aubameyang, they don’t try to get to the by-line and cross the ball, they always look for Auba making an inverted move into the penalty area.

To try to illustrate this point, I have overlaid the arrows above onto a new piece of primary data I am collecting – the below shows the ‘Turnover Tracker’ which I am using to note the position on the field where teams lost possession. This is Arsenal’s example from the match against Fulham.

45 of 76 (59%) Arsenal turnovers tracked lie along the ‘Plan A’ route while only 15 lie on the plan B route (20%).

In the red box (plan A) where Maitland-Niles, Tierney and Auba try and force their way into the Fulham box, there lies a graveyard of failed attacks, mapped out with the red crosses.

The funny thing is, although by number the vast majority of Arsenal’s attempted attacks went down the left, Whoscored.com shows most attacks for Arsenal going down the right. I believe this means that most of those moves down the left failed to materialise into something constituting an attack, whereas, by exploiting the space on the right, Willian and Bellerin were much more able to turn their much smaller number of attempts into actual attacks.

The Whoscored data is below:

Indeed, while Fulham were very effective in defending plan A (left sided attacks), all of Arsenal’s goals, including the move that led to the corner for the second, involved attacking down the right.

If we look at Fulham’s set up, we can see a similar approach (including personnel) to that used at Wembley:

Looking at Fulham’s heat maps brings out the following observations:

• You can also see the impact of Arsenal’s lopsided formation on Fulham’s heat maps. Reed is very much operating on the right, while Cairney moves across to the centre, our double pivot does not have many touches on the left, similarly Hector is often pulled to the right touchline.
• The most striking impact maybe Cavalerio, so uninvolved on the left touch line, in the first half, he frequently came to the right flank, not to swap with Kebano but to form a double team against Tierney and Maitland-Niles! Looking a Cavs heatmap it is not at all clear that he played the game on the left wing!
• Onomah’s involvement is also highly shifted to the right as Fulham combat the Arsenal set up.

The Power of the Press

Match of the Day were critical of Fulham’s pressing of the ball, but I think that is unfair, they did stop Arsenal getting the ball forward to their attacking players consistently and by my count they won more quality possession high up the field than Arsenal and, in doing so, also created two decent chances (including the early 1 on 1 for Kamara).

As another source of primary data in the game, I noted the number of times each team tried to press the other (defined as advancing players to put pressure on the opponent in their own half) as well as the outcome. These outcomes are summarised below:

The table shows what happens when each team tries to press (so first row describes Arsenal pressing against Fulham who are in possession). Fulham pressed the ball high up the field more (41 times to Arsenal’s 32 although Arsenal had the ball more so that makes sense) and had a slight edge when it came to winning the ball high up the field doing it 5 times to Arsenal’s 3 (Match of the Day had it as 10 – 3 in Arsenals favour,.

Both teams had similar levels of success in forcing the opponent to play incomplete long passes (shown in the table as ‘Played Long (turnover)’).

Then Arsenal had quite a lot more success in playing through the Fulham press doing it about 50% more frequently than Fulham. The third Arsenal goal was also an example of playing through and breaking the Fulham press.

So while Arsenal definitely did better against the press in this game than Fulham, I don’t think the difference is huge and Fulham did a commendable job making it difficult for Arsenal to advance into the attacking third of the field.

Fulham in Attack

Below we see the Turnover Tracker for Fulham (Fulham attacking right):

From this I would highlight two things:

1. Fulham lost a lot of possession on the attacking right quadrant, these failures mostly reflect occasions when they stopped Arsenal executing ‘Plan A’ and attempted to counter, they then faced the problem that, because of the lopsided Arsenal set up, the side of the field they are trying to counter through was densely packed, making progress very difficult!

2. There are a large number of x’s in the Arsenal penalty area. This is a key point, Fulham put in a good number of crosses (16) and generally got decent looking balls into the Arsenal box. Of course they never looked like getting on the end of one of those crosses or passes, but perhaps with a certain Serbian getting more minutes, that might change.  

Final Tactical Observations

To round off the tactical observations, I will reference the average player position chart from WhoScored.com below:

Here you can see the Tierney (3), Mailtand-Niles (15) and Aubameyang (14) triple up on the Arsenal left as well as the fact that they did not manage to progress much beyond the half-way line. You can also see how Fulham shifted right to respond.

Then I would note the position of Bellerin and Willian (numbers 2 and 12) on the Arsenal right, their advanced position showing that they had an easier time making progress on their flank than their three left sided teammates (likely profiting from the additional space available).

A couple other things I noticed from the game, one positive, one negative

The positive: Anguissa looked very good when he game on: he got on the ball, (around a touch a minute while on the field and most of those in the Arsenal half, no one else came close to that) and wracked up successful tackles and dribbles – indeed he was tied first in the match for successful tackles and dribbles even though on the field for just 27 minutes. That will sound slightly less impressive when I tell you that this only required 2 of each, but still, its positive! His heat map below is sparse reflecting a lack of minutes, but the location of the touches are positive being mostly in the attacking half.

The negative: When Anguissa came on, so did Mitrovic and suddenly Fulham looked like the 17/18 team again to me. For a horrible 10 minutes our press became ineffective as Mitro lumbered around after the ball and, just as teams did two years ago, Arsenal were able easily reach the final third of the pitch after encountering only minimal resistance on the way. A tactical switch followed quickly, Decordova Reid came on and it looked like Fulham went from 4-2-3-1 to 4-1-3-2 with Decordova-Reid supporting Mitro up top. Things improved a bit but then Reed looked overloaded operating on his own in the pivot. I felt we looked better defensively with Kamara on the field. Perhaps tiredness was a factor too, but this is an issue to keep an eye on in future and its an area I intend to devote some future analysis to.

Unforced Errors

One last topic which I would like to touch on, and that is the two big unforced defensive errors made by Fulham leading to goals. There is not much to say about the first one, Ream simply mis-kicked a ball and we can’t afford for that to happen.

The defending for the corner warrants looking at in more detail though, as it is really shocking.  I have taken some screen grabs from the FFCTV coverage below.

Screengrab 1: The corner is about 1 second from being taken, the eventual scorer, Gabriel, is quite hard to spot as he is hiding behind Hector. At this moment, there are 4 Arsenal players (Holding, Aubameyang, Eleny, Gabriel) in between Hector and Onomah. When the kick is taken, these Arsenal players are going to try and block all the defenders and leave Aubameyang open. And I cannot emphasise enough that at this moment, no one is marking Arsenal’s top scorer.

Screengrab 2, the moment the kick is taken. Gabriel pops in front of Hector, I believe to stop him getting to Aubameyang, Elneny is blocking Onomah and Holding goes into Rodak.

Screengrab 3: the ball is in flight: Arsenal have successfully put a ring of blockers around Aubameyang, and if the ball finds him, he is going to have an uncontested header.

Good job by Cairney and Ream putting the double team on Xhaka by the way!

Screengrab 4: The corner misses Auba, but Gabriel is ball side of Hector and also has a free header. Xhaka seems to have escaped his double team as well and is there to knock it in if either of the unmarked players ahead of him don’t get it!

In the end, we face a situation where any cross to the middle or back of the 6 yard box will be met by a free header. You can bet every team we face will watch the video of this and will look to pull a similar stunt, so lets hope Fulham get practicing marking at corners.

Conclusion

In my opinion this is a game that is about two unforced errors. Fulham were not opened up by good Arsenal play at all, the defensive pressing game was pretty good and Arsenal struggled to create chances, except when Fulham gifted them.

Because these errors were unforced, they were not, in my opinion, a function of being in the Premier League and I don’t think we will make them, at this rate, in future. I think Fulham were unfortunate and maybe unprepared on the second one.

I was also encouraged by the impact of Anguissa and would like to see him start soon. The other big positive was the amount of balls we got into the opposition box which a fully fit Mitrovic might profit from.

On the downside and on a related note, I was worried by the effect Mitrovic seemed to have on our defence, his lack of mobility (maybe fitness is an issue for him following injury) seemed to leave him unable to contribute to the otherwise effective press, It looked for a while like we were defending with a player missing. I hope this can be fixed, but I have a nagging and growing worry that this may become an important factor in our season. I hope I am wrong.

The other big negative is of course the lack of offensive production, but I am optimistic that our new signings, and the return to fitness of our striker will help in this regard.

Thank you for reading to the end! I know this is a long read and I appreciate you taking the time to follow through it, I hope you found it informative and interesting, if so please do take the time to follow me on Twitter (@Cottageanalytic) and/or let me know what you think in the comments.