Who makes the Leeds United midfield Klich?

In this piece, Tom Alderson looks at the players who have partnered Mateusz Klich this season to determine the best combination…

Six players have played alongside Mateusz Klich in midfield over Marcelo Bielsa’s two seasons at the club: Pablo Hernandez, Samu Saiz, Tyler Roberts, Adam Forshaw, Stuart Dallas and Jamie Shackleton.

In this article, I’ll take a look at what they each bring to that position and the impact this has on Klich and Leeds as a team.

Pablo Hernandez

Pablo Hernandez is the player who has partnered Klich in midfield the most since the start of last season with his minutes in central midfield increasing this season due to the arrival of Helder Costa.

SmarterScout uses algorithms to quantify a player’s actions on the pitch and how this contributes to their team’s expected goals and assists. This is adjusted for the league the player competes in.

The graphic below compares Hernandez’s stats in central midfield over the last two seasons:

From looking at Hernandez’s ‘Attacking output’ values of 99 in 2018/19 and 96 in 2019/20, we can see how outstanding he has been in central midfield for Leeds.

His style of play when partnering Klich was similar across both seasons. Hernandez attempted 2.67 shots and accrued 0.23 expected goals per 90 minutes: the highest of any of the six players to play alongside Klich. Hernandez also has a high rating for ‘Passing towards goal’ in both seasons due to the fact he has been the main creator of chances.

There are three differences between Hernandez’s two seasons. Hernandez attempted 5.43 dribbles per 90 minutes last season, compared to 3.77 this season. This is likely due to a lack of dribbling ability amongst the Leed’s squad last season meaning Hernandez had to fulfill that role in transition, as opposed to this season where Costa can fulfill the role instead. This explains the drop in Hernandez’s dribbling style rating from 67 to 29.

The other differences are due to the areas Hernandez is occupying on the pitch. Hernandez’s heatmaps for both seasons can be seen in the graphic below:

If we focus on the 2018/19 season first, Hernandez had more freedom to move around the pitch. He tended to play mostly in the inside right channel but also dropped deep and played in the inside left channel. The colour of the blocks are mainly green/yellow, meaning more medium-range and long-range passes.

This season Herandez has roamed around the pitch less but has still focused his play in the inside right channel. The colours of the blocks have changed, with the colours now blue/green signifying shorter passes. This change explains why Herandez’s ‘Link up’ rating has increased from 25 last season to 60 this season. It is the likely explanation of the increase in Herandez’s ball retention from 21 to 41.

Samu Sáiz

Samu Sáiz played the first fourteen games of last season in midfield with Klich, meaning he has played the third-most minutes in that position despite leaving in December 2018.

In the SmarterScout dashboard, we can see that Sáiz also has a high ‘Attacking output’ with a rating of 91, highlighting how good Sáiz was for Leeds before his departure.

It can be seen from looking at Sáiz’s ‘Style ratings’ that the two main features of his play were shooting and dribbling. Sáiz attempted 6.09 dribbles per 90 minutes in central midfield — the highest of any of the six players to play in that position (Hernandez was the next highest at 4.34).

If we focus on the halfway line and in the defensive half in the heatmap above, it can be seen that the colour of the blocks are blue/purple in areas. This represents Sáiz dropping deep then dribbling with the ball. Sáiz had a ground duel rating in possession of 47 which puts him below average at a Championship standard.

Looking at the colouring of the main grouping of the blocks in the heatmap, we can see that Sáiz mainly attempted medium- and long-range passes; 13% of the passes Sáiz attempted were long passes — the highest of any player to partner Klich in midfield.

However, this may have been a factor in him having the second-lowest pass completion of all the players in that position. The combination of Sáiz’s longer passing, high rate of dribbling and poor ability in ground duels may explain his low ball retention grade.

White-coloured blocks can be seen around the edge of the penalty area; representing Sáiz taking shots from distance.

Thanks to Twenty3’s Content Toolbox, we can up Sáiz’s shot map for that season:

Sáiz had the joint lowest Expected Goals per shot of any of the six midfielders we looked at with 0.06 xG which may explain why he did not score a goal whilst playing in central midfield, despite having a high shooting style rating.

Tyler Roberts

Despite never being first choice in central midfield — and it arguably not being his best position — Tyler Roberts has the second-most minutes of any player to partner Klich over the last two seasons.

Here is Roberts SmarterScout dashboard from last season:

Looking at Roberts stylistically, we can see similarities with Sáiz. In midfield Roberts has been very important to the attacking production of the side, as can be seen by his high ‘Attacking output’ rating.

The difference between him and Sáiz is the area of the pitch they operate in:

Like Sáiz, Roberts tends to drop deep looking for the ball. However, instead of dribbling, Roberts is more involved in link-up play as can be seen by the blue/green colour of the blocks around the halfway line. The majority of Roberts’ actions in the centre of the pitch is link-up play.

Across these two seasons, Roberts attempted 3.59 dribbles per 90 minutes which explains his lower dribbling ‘Style rating’. If we look at the left and right wing, we can see that Roberts tends to drift to the wings to dribble, perhaps helping to overload the play, similar to the way Helder Costa has played this season.

If we focus on the penalty area, it can be seen that Roberts was in the box much more than Sáiz, operating more as a second striker than a creative player. The main colour of the blocks in the penalty area are white meaning Roberts was mainly shooting in the box as would be expected.

Roberts shot map for last season can be seen below:

If we filter these for aerial chances, we find that Roberts managed three headed shots from a central midfield position which would explain his higher aerial style rating compared to Sáiz:

Roberts SmarterScout style ratings for the current season can be seen below:

Roberts’ ‘Shoot’, ‘Receive in box’ and ‘Aerial’ style ratings have all dropped this season with his shots per 90 minutes dipping from 2.44 to 1.57. This may be as a result of Roberts operating more as a midfielder than a second striker this season.

Comparing Roberts’ heat maps from both seasons, it appears that he’s played slightly higher up the pitch this season than last, although the similarities are remarkable:

Adam Forshaw

Forshaw is arguably Bielsa’s first choice to partner Klich in midfield. However, due to injury, he has only managed the fourth-most minutes in that position over the last two seasons. Unfortunately, due his lack of minutes no SmarterScout heat map is available for Forshaw.

Forshaw’s Style Ratings can be seen in the graphic below:

Anyone who has watched Leeds over the last two seasons will know that Forshaw is not in the side for attacking abilities like the three players already mentioned.

Unsurprisingly, then, Forshaw’s attacking output is lower than Hernandez, Sáiz and Roberts. Forshaw had the second-lowest Expected Goals per 90 minutes at 0.11 xG and the second-lowest number of shots per 90 minutes at 1.63 of the six players to play in that position.

Forshaw’s role in the side is to bring control to the midfield. He does this by linking up the play which helps Leeds to retain the ball by recycling possession which can be seen in his high ratings for ‘Link up’ and ‘Ball retention’. Forshaw’s ‘Ball retention’ rating is 35 higher than any other player in that position which shows how rarely Leeds lose the ball once Forshaw has been involved in the play.

Another feature of Forshaw’s play, especially last season, was his recoveries. Forshaw averages 8.19 recoveries per 90 minutes, nearly 3 higher than Pablo Hernandez who was the next highest. This helps Leeds defensively and may explain why Leeds had the lowest Expected Goals Against when Forshaw was in the team.

Here’s Forshaw’s Defensive Action Areas for the last two season:

Stuart Dallas

Stuart Dallas has been forced into midfield with Klich this season due to injuries to other players.

Dallas’ style ratings when playing central midfield can be seen in the graphic below. Due to his limited minutes in midfield the data may not be as indicative.

From looking at Dallas’ SmarterScout ratings, it’s hard to pinpoint what exactly Dallas brings to the team when playing in that position and may explain why he has not looked good when playing there.

Dallas only has an attacking output rating of 25 and, despite shooting being his highest style rating, he only attempted 1.7 shots per 90 minutes and accrued the lowest expected goals of the six players to play in midfield per 90 minutes at 0.1.

Dallas also has low ‘Ball retention’ and ‘Link up’ ratings despite not being a creative player which shows that he doesn’t offer the ability to retain and recycle possession in the way Forshaw does.

Dallas’ SmarterScout heat map when playing centre midfield is shown in the graphic below:

Considering Dallas is supposed to be playing alongside Klich, it was unusual to find that the majority of his play was on the wings. This may explain why Dallas was the second most common crosser of the ball of the six players mentioned, averaging 2.47 crosses per 90 minutes — Hernandez is the highest at 3.25 per 90, Roberts is third at 1.95 per 90.

If we focus on the colour of the blocks first, we can see that when Dallas does play in the centre of the pitch he plays more medium- and long-range passes, but when he moves out wide, he usually helps with link-up play through short- and medium-range passes.

Dallas may play more on the wings to help overload the play on one side but also to help out the full backs defensively. The fact that Dallas is more comfortable out wide means it makes sense to use him in this way, despite playing him lining up centrally.

Jamie Shackleton

Jamie Shackleton has played the fewest minutes so, like Dallas, his stats may not be accurate because of the small sample size.

Due to the fact that Shackleton has only started three matchings meaning the majority of his appearances are from the bench, game state must be considered when analysing Shackleton’s stats. Coming on late in the game is likely to impact the role he is asked to play.

The graphic below shows Shackleton’s SmarterScout ratings:

The two main features of Shackleton’s play stylistically according to SmarterScout are his ‘Dribble’ and ‘Link up’ play. However, Shackleton was bottom in both number of passes per 90 minutes and dribbles per 90 minutes of the players to play in that position at 29.29 passes and 1.95 dribbles respectively.

Despite Shackleton’s high ‘Link up’ rating, his ball retention rating is low compared to Forshaw who is the player most similar to Shackleton. Game state may be a factor here as later in games this season Leeds have been happier to cede possession to their opponents when leading which would likely reduce Shackleton’s ability to complete passes and retain possession.

Game state could also be the reason for Shackleton’s low attacking output. Again, if Leeds don’t need to score this will impact Shackleton’s ability to create chances and have a shot. This can be seen by Shackleton’s low ratings for shooting and passing towards goal. Shackleton only attempted 0.71 shots per 90 which is 0.92 less than the next lowest player — Adam Forshaw.

Impact on Klich

Mateusz Klich’s output changes depending on which of the six players mentioned above plays in midfield. The stats shown in the table below are Klich’s stats per 90 minutes when each player partners him in midfield:

If we look at the shots column first, the pattern seems to be that if Klich plays with attacking-minded players such as Hernandez, Roberts and Saiz, then his shot count reduces. The exception to this is Dallas, but this may be due to the small sample size or Dallas’ inability to help retain possession or create chances.

Despite Klich’s reduced number of shots when playing with Saiz and Hernandez, his Expected Goals was actually highest when playing with these players. This is likely due to their ability to create good quality chances for Klich as can be seen by the higher values in Expected Goals per shot.

Klich’s Expected Goals was also high when playing with Forshaw but this is due to him getting more lower quality chances. The data for Shackleton does not fit this pattern but this is due to the small number of minutes they have played together.

Klich’s crosses per game seem to be impacted by playing with more attacking minded players in the same way his shots are. Dallas and Forshaw produce less attacking output therefore there is more pressure on Klich to create chances which can be seen in his crossing numbers.

Finally, if we focus on the recoveries Klich completes per game; while there is no correlation between the style of play of the player partnering Klich and his recoveries, it does highlight the impact Forshaw has on Klich. Forshaw’s defensive abilities allow Klich to attack more and when he isn’t in the team, Klich has to pick up some of the defensive duties which may be the reason for his decrease in attacking output.

Impact on Leeds

Finally, we’ll look at the impact the player partnering Klich has on Leeds overall. This isn’t an exact science as attacking and defensive output isn’t completely impacted by just the one player.

For example, when Forshaw played it was mostly with Hernandez at right wing but when Hernandez played in midfield it was usually with Costa on the wing which will impact both players’ data. Opposition teams will also impact the data, mostly for players with lower minutes such as Dallas and Shackleton.

Team data depending on the player partnering Klich in midfield is shown below:

Looking at Leeds’ expected goals first, it was unusual to find that xG was not highest when Roberts, Hernandez and Saiz played but when Shackleton and Forshaw played. One possible explanation for Leeds high Expected Goals with Shackleton are limited minutes, meaning his stats aren’t as representative due to his matches being against easier opposition.

As mentioned in the previous section, Forshaw allows Klich more freedom to attack. The combination of this and Hernandez usually playing when Forshaw played could explain Leeds high expected goals when Forshaw partners Klich.

Leeds created lower quality chances when Roberts and Saiz played which can be seen in the xG per shot column. As mentioned in the individual player profiles, this may be due to Roberts and Saiz taking more long shots or their inability to create chances.

Leeds were best defensively when Forshaw played, as can be seen in the xGA and xGA per shot columns. This was expected after looking at Forshaw’s player profile. As we said, he brings the most defensively when partnering Klich.

Finally, Leeds conceded the most Expected Goals Against when Shackleton played but this may be due to either game state or his lack of minutes impacting his stats as previously mentioned. Leeds when Saiz or Shackleton played conceded the highest quality chances despite Leeds having the third-best defence out of the six players when Saiz played.

Who should partner Klich?

From looking solely at the team stats, the obvious answer to play with Klich in midfield when everyone is fit (or at the club in Saiz’s case) is Adam Forshaw.

Of the six players, Leeds ranked second for xG and xG per shot and top for xGA and xGA per shot. This led to Leeds having the best xGD when Forshaw played.

A big decision for Bielsa (fingers crossed!) may come next season when Forshaw, Costa and Harrison are all fit as there would only be two spaces available assuming Klich and Hernandez definitely play.

You can follow Tom Alderson on Twitter @tomalderson97.

Thanks to Twenty3 and Wyscout for the data visualisation.

If you enjoy this content and want to help us create more and better pieces, why not consider signing up to our Patreon page to get bonus material?

If you have any interest in contributing to this blog, get in touch with us @AllStatsArentWe.

A Leeds United blog which focuses on the tactical and statistical aspects of the game