How in the World Does Nice Overperform Expected Goals? Part One: The Defence

Part one of a two-part series. Here’s part two.

After a strong season under Claude Puel in 2016, finishing 4th, Nice got even stronger under the control of Swiss manager Lucien Favre, finishing 3rd in Ligue 1 and securing a Champions League qualifying berth in 2017. However, as a (self-professed) stats guy, I like to look at Expected Goals – soccer analytics’ most popular model – to see if the league table lies.

Ligue 1 xG Table

I know, Nice aren’t in the top 3. They’re not in the top 6 either. Not even in the top 10. Yeah, they’re 15th. Fifteenth.  How in the world did they finish 12 places above expected? This level of ‘overperformance’ is absolutely ridiculous. Here’s another table to look at the margin of overperformance:

xG Efficiency.png
Correction: everything you see as a negative number in the ‘Defensive Efficiency’ column is supposed to be a positive number.

Nice overperform xG by the greatest margin in Ligue 1. They have scored 12 goals above expected and conceded 25 below expected. Nice’s performance simply cannot be sustainable.

But then, Lucien Favre’s teams have overperformed xG before, too. As the brilliant Dustin Ward showed in his article explaining Favre’s Gladbach, Gladbach sustainably overperformed xG in three consecutive seasons (this graph was pinched from his article):

Dustin Ward Gladbach xG Overperformance.png

So Nice’s performance can’t be all luck. There has to be something about Lucien Favre’s tactical plan that makes them overperform xG and make us spreadsheet-loving nerds scratch our heads. So to explain what makes them destroy xG models, I’ve written a two-part series. And here’s part one: the defence.

The Defence

Nice have conceded the 3rd least goals in the league with 36 goals conceded, right behind Monaco (31) and PSG (27). But what if I told you they have conceded the 6th most shots in the league, with 429 shots conceded; a far cry from Monaco’s 313 and PSG’s 293.

A smart reader would say, “But shot quantity doesn’t give you the whole picture. Maybe they allow horrible shots?”

The best proxy for shot quality is the amount of Expected Goals conceded divided by the number of shots conceded. But Nice has conceded the 2nd highest xG/shot, 0.142. PSG, on the other hand, has conceded 0.101 xG per shot.

That's not very nice.gif

According to expected goals conceded, Nice have the 5th worst (or 15th best, whichever way you want to look at it) defence in the Ligue 1, with 61.09 xG conceded. With only 36 actual goals conceded, their defence overperforms by more than 25 goals. How in the world does Lucien Favre do this? Or, what exactly do they do?

  1. Nice defend with the ball.

Nice’s pass dominance (or, as Opta call it, possession) is 57% (attempted passes for/attempted passes for+ attempted passes against).  I know what you’re thinking, isn’t possession a measure of offensive dominance?

Let’s look at it this way. While playing against Nice, teams have had the ball 43% (100-57) of the time. As Johann Cruyff, the high priest of possession-based football, said, ‘You have what the opposition don’t, and therefore they can’t score.’ Possession is a defensive weapon, maybe even more than it is an offensive one, like what Louis van Gaal’s spell at Manchester United taught us. So maybe Nice avoid conceding too many good chances by keeping possession? Certainly, but it doesn’t explain how they overperform xG defensively by such a huge margin.

  1. Favre favours a low block.

Nice have conceded the most passes per defensive action in Ligue 1. It’s worth noting that this is passes and defensive actions for the entire pitch, not just a particular area.

Pressing in Ligue 1.png

However, teams that press deep generally look to avoid conceding shots from dangerous areas and don’t mind conceding shots from far out or wide. Nice, typically, do the opposite. They have conceded 46.08% of their shots from the danger zone (follow this link for more on shot zones), which is the 2nd highest in the league, and 35.25% of their shots from outside the box, lowest in the league. This is crazy.

Ligue 1 Shots Conceded Zones

Nice have also allowed the 4th highest number of key entries (passes, dribbles, turnovers, and shot rebounds resulting in possession 18 yards from goal) in Ligue 1 with 903. PSG, on the other hand, have allowed 598. So, this means Nice let opponents enter dangerous areas frequently. More on key entries next.

  1. They sit narrow, pack the central areas, and force opponents wide.

31.56% of Nice’s key entries were box entries. That’s the 6th lowest in the league. To sum up, they allow their opponents to get into dangerous areas, but most of those entries are into wide areas. You can find Nice in the fourth quadrant of this scatter.

Key Entries Against in Ligue 1.png

This is also a common feature when you watch Nice play. They put tons of players in the danger zone, which makes their opponents have to move down the wings in hope of creating chances via crosses.

  1. Opponents find it easy to get into dangerous areas against them.

It’s pretty obvious that Nice’s strategy is to invite the opposition into their defensive third (everything I’ve stated points to that), but they don’t make it hard for the opposition to get into those areas. Instead of trying to make their opponents pass the ball in Nice’s defensive third (in front of their parked bus), they let teams in and park the bus right in front of the goal. They have conceded high xG whilst not conceding too many passes. Nice have conceded the 6th highest xG/passes against.

xG Conceded Passes Against in Ligue 1.png

  1. They apply pressure on the player taking a shot.

While Nice committed only 2.53 players between the player taking the shot and the goal per shot (15th in the league), they applied the second highest defensive pressure on the player taking the shot, per shot in Ligue 1. ‘Defensive pressure’ is a subjective stat coded by StrataBet, which captures the magnitude of pressure under which the player takes a shot. It takes into account pressure exerted from all sides, not just from one direction, so there’s bound to be differences in the data. You can find Nice in the 1st quadrant of this scatter.

Defensive Pressure and Defensive Players per shot conceded

  1. They have a super keeper.

Of course, that has to be a reason. This guy,


chubby, 5’11” Yoan Cardinale saved Nice around 14 goals last season. To truly understand Cardinale’s brilliance, we have to look at (you guessed it) xG tables. This time, with a model built only on shots on target.

Raven's Ligue 1 Keeper Table

As you can see, Cardinale is by far the best keeper in Ligue 1. Top clubs should be after the 23-year-old if they want a brilliant keeper who can provide great performances for many, many years for close to nothing.

Putting it all together and looking at it tactically

I had a bit of trouble putting the pieces together and isolating a clear tactical strategy. But then, Nathan Clark, Ferdia O’Hanrahan, and Dobias Jakub made it all clear to me.

Basically, Nice try to limit the number of shots on target conceded. And like all other teams that favour a low block, Nice try to decrease the quality of shots conceded. Other teams, however, try to decrease the quality of shots conceded by focusing on avoiding conceding shots from dangerous areas. Nice, on the other hand, focus on increasing the pressure applied on the shooter to decrease shot quality. They sit back and overload some areas. It leads to the opponents having lots of time and space to play passes into dangerous areas, and that leads to lots of shots from good areas. However, the shooter finds himself in an area of high pressure, making it very hard for him to place the shot. Here’s an example of Nice applying high pressure on the shooter, this time on Ajax’s Hakim Ziyech.

Ajax v Nice Defensive Pressure.png

When the opposing team plays passes close to the Nice goal, the Nice defenders stay close to the ball but don’t look to win the ball back. They are tightly packed and sit narrow. This, however, makes them vulnerable to switched balls and through balls because full-backs and wingers who play close to the touchline are left unmarked. An example of this was when Nice played Marseille at home. A diagonal through ball from Aaron Leya Iseka reached Florian Thauvin, whose shot went into the top-right corner of the net.

Nice Defensive Weaknesses Their defense is ball-oriented and can be broken.png

This is a very passive style of defending, and it is similar to what Burnley do (and they overperform xG as well, as illustrated by Mark Thompson in his wonderful article).

So, basically, we can find one hole in most xG models: they don’t account for defensive pressure. Raven Beale, whose xG model I used for the numbers in this article (and the next), now uses Stratagem’s defensive numbers in his model. In any case, xG is here to stay, no matter what.

Se queda

Anyway, that’s all, folks, for the defensive side of it. As for Nice’s attack, I’ll post the article in a week. Hopefully.

This article was written with the aid of StrataData, which is property of Stratagem Technologies. StrataData powers the StrataBet Sports Trading Platform, in addition to StrataBet Premium Recommendations.

I’d like to thank Raven Beale, Nathan Clark, Dobias Jakub, Ferdia O’Hanrahan, @ashfutbol7, @Sarriball, @clincalbailly, Martin Hawkes-Teeter, and Stewart Brown for their help with this article. I couldn’t have written this without your help.

Hope you enjoyed the article. If you liked it, don’t forget to share it with others. If you have any feedback/questions regarding this article, feel free to use the comments section below or find me at Twitter at @thefutebolist.


4 thoughts on “How in the World Does Nice Overperform Expected Goals? Part One: The Defence”

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s