Millwall FC Case Study
Millwall 2018/19 season analysis
Millwall lined up in either a 4-4-2 or a 4-5-1 formation.
The average pass position of each player with over 300 passes over the season is shown below.
The size of the dots represents the number of passes each player made.
You can see that the players on the right-hand side are more involved in build-up play than those on the left.
The difference between the Millwall style and that of Swansea is the positioning of players when they make their passes, and the relative equality between each of the players.
Swansea play many more passes between their players in their own half of the pitch seeking to dominate the ball.
Millwall play more directly, playing relatively long passes.
Looking at the passing networks (we have created these ourselves using data recorded from every Championship match) we can see that Millwall favour passes from central areas into wide areas. And again, slightly favour these coming in from the right-hand side of the pitch.
By analysing every pass and assigning it a score using our Pass Score system we can see where teams create their danger from.
The Pass Score system uses 1.2m individual passes played in the Championship over the last 3 seasons.
These are all the pass starting locations played by Millwall that scored highly on our system. It shows the majority of good passes were played from wide attacking areas, relatively few in central areas.
If we look where these passes were played to we can see that the vast majority were passes into the penalty box from wide areas. The high scoring passes in wide areas are likely to be successful longer passes into the wide areas. These channel passes are a feature of the Millwall game plan.
We can further break this down by looking at the types of passes each player makes and their how highly we rate these passes in terms of creating goalscoring chances.
So we can see that Cooper and Hutchison are attempting passes that generally have a low success rate (long passes deep into opposition territory) and these are not creating dangerous attacking situations.
The midfield trio of Williams, Tunnicliffe, and Leonard play easier than average passes and are not creating chances. Again this could be tactical, they may not try and create chances directly, instead pushing the ball into wide areas or retaining possession.
Gregory was particularly good at knitting play together with well above average quality of simple passes that create chances whilst the Millwall wide players, particularly Ferguson and Wallace show excellent ability to play difficult passes that result in chances.
Again Millwall’s tactics show up clearly with wide players and the striker the dominant creators.
When we compare all the Millwall players to the overall league-wide level of “expected passes” we can see that almost every player is completing passes at a below-average level. Only Tunnicliffe completes more passes than would be expected. This may indicate a tactical preference for territory but could also indicate areas where player quality could be improved. Again the dominance of the right-hand side shows up with Wallace and Romeo showing up as the highest volume passers.
A perfectly balanced team would show up with very little difference between whether passes were played left or right. The below diagram looks at which sides teams favour passing to as they progress up the pitch. With Bristol City we can see the generally aim for the left hand side of the pitch to build their attacks. With Millwall we can see they favour the right.
We can also see that Millwall are one of the lowest volume passing teams in the league from deeper areas but actually still have as much passing activity in the final 10 yards closest to the opponent’s goal as other teams.
The above diagram shows passes played from their own defensive third. It shows teams who favour longer passes at the botttom left and shorter passes top right. Teams below the dotted line play their passes at below-average levels of completion.
The below diagram shows passes into the final third. Bottom left play longer more difficult passes, top right easier passers.
Important players
It is no surprise that Millwall play longer than average passes, create more chances from crossing and favour their right-hand side whilst building attacks.
However by analysing their preferred style of play we can:
Identify the player’s positional attributes
Look at the positions where player quality could be upgraded
Identify players with similar attributes but with the potential to offer more quality.
Identify leagues where we are likely to find more of these players to match the qualities demanded, so resources can be targeted better.
Milwall 2018/19
Player profiles: Glossary of terms
All templates compare player performance to the median (average) player in their position, in their league.
Some of these statistics indicate playing style rather than player ability.
Centre back template
Passing and Progression
Final third to Forward Pass ratio: A low bar indicates they favour short forward passes into midfield rather than longer passes into the final third
Final third pass accuracy %: A low bar indicates low accuracy passes into the final third of the pitch
Fwd pass ratio: A low bar indicates they play more sideways and backward passes than the average player in their position
Fwd pass accuracy %: A low bar indicates low passing accuracy for forward passes
Passes per 90: A low bar indicates a lower number of passes per game. Excludes crosses.
Pass accuracy: A low bar indicates inaccurate passing
Long passes accuracy %: A low bar indicates inaccurate low passing
Long pass ratio: A low bar indicates a player who plays relatively few long passes
Average Short/Medium pass length: A low bar indicates shorter than average pass length
Average Long pass length: A low bar indicates shorter than average long pass length
Creativity and Attacking Play
Dribbles per 90: A low bar indicates fewer attempted dribbles per 90
Dribbles success %: A low bar indicates a low success rate at attempted dribbles
Deep completed passes per 90: A low bar indicates fewer successful passes that finished within 20 metres of the opponent’s goal
Smart passes per 90: A low bar indicates fewer “smart” passes per 90. Defined by Wyscout as a pass that develops the attack in a creative way
Smart pass accuracy % -A low bar indicates a low success rate on smart passes
Thru passes per 90 – A low bar indicates fewer passes into space behind a defensive line
Thru pass accuracy % – A low bar indicates low thru pass accuracy
Key passes per 90 – A low bar indicates fewer passes that lead directly to a shot
Passes to penalty area per 90 -A low bar indicates fewer passes into the penalty area
Defensive performance
Aerial duels per 90: A low bar indicates the player is involved in fewer aerial challenges
Aerial duels won %: A low bar indicates a lower success rate for aerial duels
Defensive duels per 90: A low bar indicates the player is involved in relatively few defensive duels
Def duels won %: A low bar indicates a lower success rate for defensive duels
Fouls per 90: A low bar indicates a player who gives away more freekicks
PAdj Interceptions per 90: A low bar indicates a player who does not intercept the ball as frequently. This is adjusted for possession.
PAdj Tackles per 90: A low bar indicates relatively few tackles per 90. This is adjusted for possession.
Tackle success %: A low bar indicates a low tackles success rate
Shots blocked per 90: A low bar indicates fewer shots blocked than typical.
Full back template
Passing and Progression
Same as CB Template
Attacking and Creativity
Cross to Pass Ratio: A low bar indicates they make more passes for each cross than average
Cross %: A low bar indicates a low cross success rate
Dribbles: A low bar indicates they do not attempt to dribble frequently
Dribbles Success %: A low bar indicates a low success rate for attempted dribbles
Touches in box: A low bar indicates a low number of touches in the opposition box
xA: A low bar indicates they are creating fewer good goalscoring chances (expected assists)
Succ. through balls: A low bar indicates fewer successful passes beyond defensive line
Succ. Smart passes: A low bar indicates fewer successful creative attacking passes
Succ. passes to penalty area: A low bar indicates fewer successful passes into opposition box
Defensive performance
Same as CB template
Midfielder template
Passing and Progression
Same as CB Template
Attacking and Creativity
Succ. Dribbles p90: A low bar indicates relatively few successful dribbles
Deep completed passes per 90: A low bar indicates fewer successful passes to within 20 metres of the opposition goal
Thru passes per 90: A low bar indicates fewer passes into space behind opposition line
Thru pass acc. %: A low bar indicates low accuracy of passes into space behind opposition
Succ. Smart passes: A low bar indicates fewer successful creative attacking passes
xA per 90: A low bar indicates they are creating fewer good goalscoring chances (expected assists)
Passes to penalty area per 90: A low bar indicates fewer successful passes into opposition box
Touches in box per 90: A low bar indicates fewer touches of the ball in the opposition box
Shots per 90: A low bar indicates fewer attempted shots
NPxG per 90: A low bar indicates low levels of non-penalty expected goals (low goalscoring threat)
Defensive Performance
Aerial duels per 90: A low bar indicates the player is involved in fewer aerial challenges
Aerial duels won %: A low bar indicates a lower success rate for aerial duels
Defensive duels per 90: A low bar indicates the player is involved in relatively few defensive duels
Def duels won %: A low bar indicates a lower success rate for defensive duels
Fouls per 90: A low bar indicates a player who gives away more freekicks
PAdj Tackles per 90: A low bar indicates relatively few tackles per 90. This is adjusted for possession.
Tackle success %: A low bar indicates a low tackles success rate
PAdj Interceptions per 90: A low bar indicates a player who does not intercept the ball as frequently. This is adjusted for possession.
Attacker template
Passing and Creativity
Succ. Smart passes: A low bar indicates fewer successful creative attacking passes
Succ. Passes to penalty area: A low bar indicates fewer successful passes into opposition box
Succ. through passes: A low bar indicates fewer succesful passes into space behind opposition line
Deep Completions: A low bar indicates few successful carries or passes of the ball to within 20 metres of the oppositon goal
Final third to Forward Pass ratio: A low bar indicates they operate in deeper areas playing more forward passes that do not reach the final third than average
Final third pass accuracy %: A low bar indicates low accuracy of passes played in the final third
Passes: A low bar indicates a lower number of passes per game. Excludes crosses.
Pass accuracy: A low bar indicates inaccurate passing
Avg Pass Length: A low bar indicates shorter than average passing
Attacking Play
Cross to Pass Ratio: A low bar indicates they make more passes for each cross than average
Cross %: A low bar indicates a low cross success rate
Dribbles: A low bar indicates they do not attempt to dribble frequently
Dribbles Success %: A low bar indicates a low success rate for attempted dribbles
Goal build up: A low bar indicates low involvement in either the assist, the pass before the assist or the pass before that.
xA: A low bar indicates they are creating fewer good goalscoring chances (expected assists)
Shots: A low bar indicates fewer attempted shots
xG/Shot: A low bar indicates a lower average quality of chance per shot
Touches in box: A low bar indicates fewer touches of the ball in the opposition box
Defensive Performance
Same as Midfielder template
Striker template
Build up play and creativity
xA: A low bar indicates they are creating fewer good goalscoring chances (expected assists)
Goal build up: A low bar indicates low involvement in either the assist, the pass before the assist or the pass before that.
Succ. Smart passes: A low bar indicates fewer successful creative attacking passes
Succ. Passes to penalty area: A low bar indicates fewer successful passes into opposition box
Deep Completions: A low bar indicates few successful carries or passes of the ball to within 20 metres of the oppositon goal
Final third pass accuracy %: A low bar indicates low accuracy of passes played in the final third
Pass accuracy: A low bar indicates inaccurate passing
Passes: A low bar indicates a lower number of passes per game. This excludes crosses.
Avg Pass Length: A low bar indicates shorter than average passing
Shooting and attacking play
NP Goals: A low bar indicates fewer non-penalty goals
NP xG: A low bar indicates fewer non-penalty expected goals
xG Performance: A low bar indicates they are scoring fewer goals than expected for the quality of chances they have had.
Headed goals: A low bar indicates fewer headed goals scored per 90
xG/Shot: A low bar indicates a lower average quality of chance per shot
Shots: A low bar indicates fewer shots per 90
Shots on target %: A low bar indicates less accurate shooting
Goal Conv %: A low bar indicates fewer shots leading to goals
Succ. Dribbles: A low bar indicates fewer successful dribbles
Touches in box: A low bar indicates fewer touches of the ball in the opposition box
Defensive Performance
Same as Midfielder template
Squad changes
Players departing played a combined 11,166 minutes:
Lee Gregory (3636)
Steve Morison (1976)
Jordan Archer (2160)
Ryan Tunnicliffe (1958)
David Martin (900)
Conor McLaughlin (536)
The positions that need replacing are goalkeeper, central midfield and attack.
Signings made to this point are;
Matt Smith – QPR – Striker
Jon Dadi Bidvardsson – Reading – Striker
Frank Fielding – Goalkeeper
Connor Mahoney – Wide attack
Alex Pearce has also joined on a permanent deal.
Mahoney, Smith, and Bodvarsson all fit into the club’s preferred game plan with Mahoney excelling at crossing and dribbling. Smith and Bodvarsson are excellent aerially.
With the club fully committed to a style of play it is important that player recommendations take this on board.
We, therefore, build a template for each position in the squad.
This shows the types of qualities we need from each player position.
There is flexibility within this. There is no point in finding exact replicas for existing players. You ideally want players who are capable of fulfilling the tactical requirements of the team but also adding qualities.
An example of a club who maintained the 4-4-1-1 shape whilst improving the quality of the individual players was Everton under David Moyes. Initially, they played with a flat back 4, a flat midfield that attacked mainly from wide areas and a target man forward. However over time adding players like Arteta, Baines and Pienaar allowed the team to maintain the traditional English qualities and shape but with additional ball retention higher up the pitch. They would still aim direct passes to Fellaini but also had the option of waiting for a better crossing opportunity with Baines and Pienaar able to keep the ball between them in the final third.
Areas to strengthen – Undervalued players
Players in this section are recommended based on video scouting and have been assessed as fitting the criteria to work successfully in a 4-4-1-1 formation playing direct football. They are also selected as players likely to increase in value.
Goalkeeper
Millwall released 3 goalkeepers in the summer (Archer, Martin, and King) bringing in Frank Fielding.
Goalkeeper has traditionally been a difficult position to assess with data. This has changed recently with improved shot-stopping metrics from Statsbomb and Opta. What they don’t assess as well as the eye is the style of distribution and ability to cope with pressure on high crosses.
This is why in addition to data we recommend recording actions yourself to create your own data. Look at approximately 150 distributions and 150 crosses per goalkeeper and record whether they performed as you would like your goalkeeper to perform.
We would recommend Denmark as being a league where goalkeepers are regularly tested with aerial balls, aggressive set pieces are common and direct football is often played.
One goalkeeper worth looking at in particular is Jeppe Hojbjerg of Espjerg. He is a very good quality shot-stopper, excellent on crosses and with accurate distribution.
Left-back
James Meridith (31) is the only specialist left back in the squad. He is likely to play behind Connor Mahoney.
An option who would be first-team ready almost immediately is Liberato Cacace, a New Zealand U20 international playing for Wellington Phoenix. He has played the most outfield minutes of any player in his team and has been consistent in defence and attack. At just 18 years of age he is physically strong. He can overlap or underlap the winger in front of him and play accurate channel passes.
He holds an Italian passport so would qualify for a move to Europe.
Centre midfield
The attribute required for a Millwall midfielder are positional discipline, the ability to play passes to wide areas and quickly break up opposition attacks.
In order to increase the quality of the midfield, you need players who can produce the same level of defensive output with above average ability to play accurate passes.
One of the best divisions to scout in for central midfielders with this profile is France. Ligue 2 players often move for fees of around £1m. One player with the attributes to succeed at Millwall is Julien Masson of Valenciennes. He is a 21 year old central midfielder who is both a tough tackler but also has an excellent passing profile with the ability to both play passes out hiinto the space behind full-backs but also to play more difficult passes into central areas.
A £1.2m move to Angers fell through after problems with the medical due to a short term knee injury. However, with only 1 year left on his contract he may be available for an even lower fee in the short term future.
Linking attacker
Losing Lee Gregory means the squad lacks a link-up attacker. It may be possible for an existing squad member such as O’Brien to take up the role. If that is not possible there are potentially affordable players in Scandanavia who would perform this role.
Patrik Karlsson Lagemyr is a forward for IFK Goteborg. The Swedish league is quite direct and he tends to play a link role between midfield and attack.
A different option would be Daleho Irandust, a player for BK Hacken, in Sweden. He is a physically stronger player than Lagemyr but plays in a similar role.
Both players offer creative passing and the movement required to play as runners playing off a target man.
With additional, refined, requirements for the role, a larger list of affordable players could be produced.
Data Scouting recommendations
We can also recommend players based on data searches. We are able to search on any attribute a club desires, the system is completely flexible and no two clubs will have the exact same requirements.
Full back
As press reports had linked Millwall with Jason McCarthy we decided to profile him in the report.
A player with similar qualities, playing on the left hand side of defensive is Christopher Operi of Chateauroux in Ligue 2.
He is physically strong, a very good one on one defender and progresses the ball up the pitch well.
He is likely to fall within the wage structure at Millwall and has the potential to increase in value.
Midfield
For leagues we have passing data for we can create visuals that allow us to compare passing profiles. In this example we look at a current midfielder, Shaun Williams and try to find a player with similar skills who plays passes into areas that would benefit the tactical plans of the club.
Looking through our database containing every Championship pass for the past 3 years we can pick out similar passers.
Joe Williams (Everton, on loan at Bolton) passed from similar areas, just with a higher volume of good passes.
Link attacker
With this position we looked at players who were able to play in a direct system. It was deemed important that they were able to win aerial challenges but also play good passes to wide areas.
Robert Zulj profiles like this with good creative and attacking numbers but also robust and aerially dominant.
Leave a comment