Looking Into Leicester’s Hire Of Claude Puel

After an 8th place finish and League Cup final in his first season at Southampton, it seemed as though Claude Puel was unlucky to lose his job with the Saints. A poor end to the league campaign saw them only win 1 of their last 8 and go 5 home matches without scoring. Southampton decided one season was enough for Puel and hired Argentinian coach Mauricio Pellegrino.

Puel now finds himself back in the Premier League with the 2015-16 Champions. Interim-turned-permanent manager Craig Shakespeare got off to a great start with the Foxes, but was sacked after the club started the current season slowly, eventually finding themselves in 18th place.

5 Premier League clubs have now opted for a change of manager during this season, and while the returns of Sam Allardyce, David Moyes, Alan Pardew and Roy Hodgson seems to have made bigger headlines this season, it’s Claude Puel’s hiring which I think has some interesting points.

Before continuing it’s worth pointing out I’m not overly familiar with the stats behind Puel’s sides in France, so I unfortunately only have his previous season with Southampton to judge him on. Another quick note, the StrataBet data is up to date as of Gameweek 12 (21/11/2017) while my other data is up to date as of Gameweek 14 (01/12/2017).

Puel at Southampton

Stylistically, Puel seems an interesting choice for Leicester. Using the below graph (as I always do) to try to get a quick idea on a teams play style shows this:

Click to Enlarge

You can see Leicester were more direct and less active when out of possession than Southampton last season. It’s not a huge difference, given neither side were the extremes of those two measures, but it is a difference. It’ll be interesting to see if Puel tries to move Leicester into the direction of his Southampton side on the above graph, or whether there’ll be a trade-off given the players he has at his disposal and what they’re already used to – particularly until he gets a pre-season with them.

The good news for Leicester is that Puel had an organised defence when at Southampton. Their xG against of 0.97 per game was the 6th highest in the division, while using StrataBet measures shows them to have the 2nd highest average number of players behind the ball when the opposition shoots (only behind Burnley) and the 3rd highest ‘defensive pressure’ on the shooter. This culminated in Southampton having the 2nd lowest xG per chance conceded in 2016-17.

What’s interesting is when I previously looked at how players behind the ball and defensive pressure affected xG, teams tended to have either bodies behind the ball or put a lot of the pressure on the shooter. Southampton excelling at both is really impressive.

In fact if you add the two averages together no club in Europe’s top five leagues had a higher value. While their xG per chance conceded was the 10th lowest in Europe’s top five leagues, this almost makes it seem like they were the most difficult to team to have a chance against – as you’re most likely to be under heavy pressure and have lots of bodies in front of you.

The problems for Southampton last season though were in front of goal. While their xG of 1.25 per game was the 8th most in the Premier League, they only managed 0.921 goals per game (open play shots only). Their bad luck in front of goal gave them an xG Rating of 0.74, which was the 3rd worst in the Premier League and 9th worst in Europe’s top five leagues.

What makes matters worse for Southampton is that despite their healthy xG totals they had the 4th worst xG per chance last season. This seems to point towards them having lots of low to mid quality shots, with little high quality opportunities.

Reinforcing Southampton being well organised in defence and tough to play against last season is their xP against rating. While not far under 1 their 0.989 was the 6th lowest in the division. Equally impressive is that Southampton had the 4th lowest value for opposition pass completion for both the defensive (Southampton’s attacking) and middle third last season, as well as the 7th lowest for the attacking (Southampton’s defensive) third.

Unfortunately, I can’t compare the xP values across 2016-17 and 2017-18, as they use a different method and I haven’t had chance to roll the newer one out across other seasons. So the xP stuff will just compare to other teams in the division for the same season.

Leicester in 2017-18

Leicester have had a huge change in their chance numbers this year – including both Shakespeare’s and Puel’s time in charge. Last season Leicester found themselves with the lowest xG per chance in the division. So far this season, using StrataBet data again, they have the highest xG per chance with 0.177.  This is a pretty huge change for the Foxes, but what’s caused it?

What makes this change even more interesting is Leicester are having less chances per game, 7.5 down from 8.76, yet a higher xG per game, 1.329 up from 0.96. Digging a bit deeper into how they create their chances led to some interesting findings.

In my piece looking at how the action preceding the chance effects the xG value I found that low crosses were an effective way of creating higher probability chances, particularly in comparison to high crosses. Leicester seem to have reaped the rewards of this.

In 2016-17, Leciester created 0.263 chances per game from low crosses, the 2nd fewest in the division. These chances accounted for just 5.5% of their total xG. This season however they’ve created 0.833 chances from low crosses per game, behind only Tottenham and Manchester City, leading to chances that have accounted for 22.4% of their xG. On the flip side their chances created from high crosses per game is down from 0.868 to 0.583, meaning the percentage of their xG from high crosses has come down from 13.6% to just 5.6%.

It’ll be interesting to see whether this is just a fluke that’ll even out across the course of the season or some kind of conscious decision behind scenes.

While it may be preferable to try and generate more chances, 1.329 open play xG is a respectable total for Leicester. Adding in penalties (with a flat rate of 0.76 for all penalties) sees this rise to 1.519, while adding in penalties and dangerous moments increases it further to 1.713. Open play only has them with the 9th highest value while adding in penalties and both penalties and dangerous moments has them with the 7th highest xG per game in the league,

The problem however is at the other end, Leicester’s 1.645 open play xG against per game is the 4th worst in the division. They’re restricting the opposition to low quality chances with the 3rd lowest xG per chance in the division but they’re also conceding the most chances per game in the division.

Searching further into where these chances are coming from becomes interesting too. Looking strictly at location (splitting the pitch into 3 with the boundaries being the mid-point of the 6 yard box and the 18 yard box) and Leicester concede 26% of their xG from their right, above the league average of 22%. Breaking down the types of chances and it’s high crosses and open play passes in which a higher percentage of their xG against has come from when compared to the league average.

Below you can see the chances they’ve conceded via these methods:

Click to Enlarge

A high portion of the chances they concede seem to be from getting into good positions between the right-back and right centre-back, usually Danny Simpson and Wes Morgan respectively. The cluster of open play passes inside the area is particularly worrying.

After looking at a few clips it seems both Simpson and Morgan have a tendency to be drawn towards the ball, whether it’ll be them getting dragged out wide and creating space in the middle, or being incredibly narrow and letting a wide player enter the box. Some clips just from the Arsenal and Manchester United matches can be seen below:

This weakness on the right is shown even more when looking at StrataBet’s ‘Key Entry’ data. These entries record the first time a team enters the box in an attack – as opposed to all box entries.

Leicester have the highest number of opposition key entries for their right side of defence with 10.917 a game, quite a bit more than the next highest 8.417. 40.4% of their total key entries against come from their right hand side, only Southampton allow a higher percentage (41.5%).

What makes it more concerning however is that Leicester also allow the highest number of key entries in the division with 27 per game.

This points towards it being easier to enter the Leicester box per possession/attack than any other side in the Premier League. 27 entries a game means that on average every 3 minutes 20 seconds the opposition is entering the Leicester for the first time in a new attack. The weakness to the right side and giving up such a high number of key entries are things that Puel should look to address if he wants to tighten the Foxes defence.

Puel will also have to improve the passing statistics of this Leicester side. After reading this piece on StatsBomb by @WillTGM I wanted to replicate the ‘disruption’ for each zone of the pitch using the opposition xP numbers, while also looking into implementing something like @FootballFactman’s new method of measuring pressing from this piece.

Looking at the ‘disruption’ (real pass completion percentage – expected pass completion percentage) proved interesting for Leicester this season. Below shows their disruption per quarter of the pitch:

Click to Enlarge

As you can see, in the opposition’s own half Leicester do an okay job. The opposition complete passes pretty much at exactly the same rate as they’re expected to. The problem gets worse as the opposition gets closer to Leicester’s goal however. This goes hand in hand with the amount of box entries Leicester’s opposition rack up against them. Completing more passes than expected close to the opposition goal is likely to involve lots of low probability passes into dangerous areas being completed.

After looking at this I thought it’d be interesting to look disruption for vertical zones, splitting the pitch into five rows. This gave the following result:

Click to Enlarge

Again their weakness on the right is plain to see.

On the left they’re average, to allowing slightly more than expected in the left half space, whereas on the right they allow more than expected in both the half space and the wing. If Leicester want to improve their defence they need to make it harder for the oppositon to complete passes – whether it’d be through implementing a higher press like Puel’s Southampton or just fine tuning their mid to low block.

It’s worth pointing out the centre wasn’t included in this diagram above as the numbers are off the scale. I put this down to something being wrong with the model more than anything as the central disruption for Leicester was a crazy 30.8%. This may be blown out of proportion due to long-balls from the goalkeeper or players winning headers from goalkeeper kicks but it is something I’ll have to look into as a league average disruption of 12% for the centre certainly doesn’t feel right.

On the ball Leicester also experience problems with passing. Creating a similar diagram as the horizonal disruption but this time for Leicester’s xP Rating in each zone shows them under performing their expected passes in every zone:

Click to Enlarge


Puel seems like a good hire for Leicester, the defensive and passing areas where Leicester are currently weak are both areas in which Puel’s Southampton were strong. If the Frenchman can replicate his strong defensive numbers with Leicester – without sacrificing their respectable xG for – then it could be a good season for the Foxes.

It’ll be interesting to follow his progress as Leicester manager, to see if he changes Leicester’s style and to see whether or not this is a short-term fix for Leicester or they’re trying to push their style in a different direction.

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.*

*Chance and key entry data from StrataData, not the passing data.

Leave a Reply

Your email address will not be published. Required fields are marked *