How in the World Does Nice Overperform Expected Goals? Part Two: The Attack

This is part two of a two-part series explaining how Lucien Favre’s Nice overperform expected goals, that cool new stat they show on MOTD, by a huge margin. Part one explained how Nice’s defence overperform expected goals against (xGA). In part two, I explain how Nice’s attack overperform expected goals for (xGF). Let’s jump right in.

*      *      *

With 63 goals, Nice had the 4th-best attack in Ligue 1 last season in terms of goals. This is way behind the best attack, the marvellous Monaco’s, which scored 107 (it also overperformed xG by a giant margin; we’ll get to that later).

But once again, Nice’s shot tally is terrible compared to the other elite teams in Ligue 1. They took 392 shots last season, which was only 8th-best in the league and a long way off from PSG’s 516 and Monaco’s 464. Another weird thing is that their shot quality was bad too. Nice’s xG/shot score was a little less than 0.13, quite distant from Monaco’s 0.17 and PSG’s 0.16.

Looking at all of these stats together makes it a lot clearer:

Ligue 1 xGF Table.png

In this table, it’s not Nice that stands out. It’s everyone’s favourite team last season, the young, entertaining Monaco. They not only scored the most goals, but their attack also overperformed xGF by the greatest margin in Ligue 1, which helped them wrestle the Ligue 1 title away from big, bad PSG. “Why is this article about Nice then?”  you ask. Well, firstly, I already started a series about Nice and I ain’t starting again. I’m lazy. Another thing is that nothing in the data points to Monaco repeating this in the 2017-18 season. Additionally, Monaco lost a large chunk of their talented, young squad with the departures of Kylian Mbappé, Bernardo Silva, Tiemoué Bakayoko, and Benjamin Mendy. They might just be another Leicester, although a better version, and reversion to the mean looks very probable. It’s different for Nice because:

  • Favre’s Gladbach did this too.
  • There are many things in the data that point to Nice doing this (or at least some of it) again.

What exactly are the things that Nice do with the ball?

  1. Nice keep hold of the ball.

As I said in part one, Nice kept the ball 57% of the time, 2nd-most in the league. They, however, had the fewest number of possessions in the league, but when they got the ball, they kept it for 18.5 seconds on average. You can find Nice in the second quadrant of this scatter.

Possession in Ligue 1.png

  1. They prefer passing the ball out from the back and play lots of passes when they attack.

Nice played 600.1 passes per game last season, which was 2nd in the league, after PSG. They also completed 87.8% of their passes, 2nd highest in the league, after (*yawn*) PSG. This means Nice play a lot of passes in their own defensive third and in the middle third, or at least that’s what the video tells me. The guy who conducts their build-up play is the intelligent Ivorian deep-lying playmaker Jean-Michaël Seri. He played 85.4 passes per 90 minutes in Ligue 1 last season and completed 89.5% of his attempted passes. His final product was great too – he played 2.2 key passes p90 last season.

Nice played 9.8 passes per goal-scoring possession (2nd in the league, no prizes for guessing who’s first). Another thing that reflects Nice’s slow-moving, pass-heavy build-up style is the fact that they put up least xG divided by total passes in Ligue 1. They like passing the ball around a lot before getting into dangerous areas.

xG For Passes For Ligue 1.png

All this means Nice use an ultra-low block but don’t counter-attack when they get the ball. That’s not something we see often.

  1. They attack centrally.

While Nice rarely enter the opposition 18-yard area (the whole area, not just the box), they usually enter the box when they do. Stratagem calls an entry into the final 18 yards a key entry. A high percentage of Nice’s key entries are box entries. Nice finds itself in the 2nd quadrant of this scatter.

Key Entries in Ligue 1.png

Here’s more on Nice’s key entries:

  1. They make use of the few times they get into dangerous areas.

Again, Nice’s key entry tally was only 653 last season, 2nd-lowest in the league. However, 31.24% of their key entries resulted in a shot from inside the box, 2nd-best in the league. Nice are once again in the 2nd quadrant of this scatter.

Key Entries Efficiency.png

  1. They take lots of long shots.

44% of Nice’s shots came from outside the box last season, and that’s 3rd in the league. The average team in Ligue 1 would have taken 40.35% of their shots from outside the box. And what about PSG and Monaco? PSG took 32.45% of their shots from outside the box while Monaco took 33%. Those are the two lowest numbers in the league. This is the thing that makes Nice bizarre. Generally, long shots should be avoided, as only ~3% of all shots from outside the box are converted. Or at least that’s what we think. Maybe Nice do something differently? We’ll get to that later.

  1. They take their shots while under a high amount of pressure from opponents.

As we explored in part one, Nice’s defenders limit their opponents’ shot quality not by making them shoot from far off, but by applying pressure on the shooter. The fact that most xG models (including ours – I mean, Raven Beale and Peter McKeever’s) don’t account for actual defensive pressure but use proxies is the reason quality defences like Nice’s overperform xGA.

However, Nice take their shots under a lot of pressure. According to Stratagem’s ‘defensive pressure’ rating that estimates the amount of defensive pressure on the shooter on a scale of 1 to 5, Nice take their shots with 2.6 defensive pressure on average. This number is the highest in the league. So defensive pressure doesn’t explain Nice. But…

  1. They shoot when there are only a few defenders between the ball and the goal.

On average, Nice take their shots when there are 2.27 defenders between the ball and the goal. This number is the lowest in the league. The highest is just 2.63 for Bastia, but there’s still a large difference when you look at individual shots. 59.13% of Nice’s shots were taken when there were 2 or fewer defenders between the ball and the goal, and this number is the third-highest in the league. On the other hand, only 7.2% of Nice’s shots were taken when there were 5 or more defenders in between the ball and the goal (I only looked at open play shots, as penalties, free-kicks or corners would heavily distort the numbers).

I think this is the biggest factor that explains Nice’s attacking efficiency. It’s easier to take better shots when you have fewer defenders between you and the goal, right? Another thing: on average, only 2.8 defenders block the shooter for Nice when it comes to shots from outside the box. That number is the lowest in the league. This is probably why they take as many long shots as they do.

Putting it all together and looking at it tactically

Nice, as we discussed in part one, use an ultra-low block. Their opponents can take many shots from dangerous locations, but they have to deal with a high amount of pressure when they try to get their shots away.

Usually, teams that use a deep press launch swift counter-attacks when they get the ball back. Nice, however, get the ball back either by absorbing shots (23.5% of Nice’s possessions sprang from absorbing shots, 3rd highest in the league, significantly higher than PSG and Monaco) or through last-ditch tackles or interceptions, making it hard to start counters.

So, Nice play the ball out from the back and work slowly towards the opposition goal. Of course, if you keep possession and play lots of passes, you will face a low block. From their video, it’s clear that they play many passes in good areas, presumably to pull the opponent’s ‘parked bus’ out of shape. And when the opponent tries to readjust, Nice shoot, irrespective of location. This explains the low number of defenders between the ball and the goal when Nice shoot. An example of this was when football’s bad boy Mario Balotelli scored against FC Krasnodar in the Europa League:

Here’s another example, Anastasios Donis’s long shot against Schalke in the Europa League:

Sometimes, Nice do counter-attack, generally against teams that press high and keep the ball. It can lead to something that’s nothing but pure art.

In the end

It’s important to note that defensive data doesn’t explain all of Nice’s overperformance. Some of it can still be luck.

Another thing: the fact that Lucien Favre’s teams overperform xG does in no way mean that xG is useless. It just means we have to find ways to account for defensive pressure and defensive players between the ball and the goal. This is easy with data from Stratagem, where you literally get those two subjective ratings in a neat spreadsheet, but it’s harder with Opta data. Sure, guys like the amazing Michael Caley, the god of xG on Twitter, use tons of proxies for defensive pressure, but it’s still not enough. But here are some things that, in my opinion, can be added to improve xG models:

  1. A reduction in xG if a shot is blocked. This was nabbed from Raven Beale, who used it in his earlier models.
  2. Adding an adjustment for passing in dangerous areas in the possession chain that led to the shot.
  3. League/team effects. I know xG is supposed to be built on averages, but some teams apply more pressure on the shooter than others, and it’s worth adjusting for.
  4. Studying what exactly happened in the possession that led to the shot (thank you, Good Doc). How did the attack start? How many passes did the attack consist of? How fast was the attack? Will Gürpinar-Morgan’s OptaPro Forum presentation provides a great starting point.

For a stat that just gained mainstream recognition (even if it’s two years after it first became popular among stat nerds), it’s also important to note that xG isn’t always correct. Similarly, there can be no dogmas related to this. For example, long shots. Are long shots necessarily a bad strategy?

Uh, well, they are.

But we need context. How much space did the shooter have when he took it? It’s easier to shoot from 20 yards out with a fairly low amount of pressure compared to 5 yards out with 5 players and the keeper trying to get the ball back.

But one thing’s for sure, xG is here to stay.


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.


InStat_logo transparent

The statistical information to this article was provided by InStat. InStat is a world leader in sport statistics, cooperating with more than 1300 football clubs all over the world, including such giants as Chelsea, Barcelona and Bayern Munich. InStat provides its users with a wide range of football statistics products, starting from actions reports to video platform InStat Scout.

For collaboration requests contact:

ruslan.vilikiev@instatsport.com


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.

Advertisements

One thought on “How in the World Does Nice Overperform Expected Goals? Part Two: The Attack”

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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 )

Google+ photo

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

Connecting to %s