What Does the Data Tell Us About Leeds at Set Pieces?
In this piece, Tom Alderson investigates the numbers behind Leeds United’s set pieces from the 20/21 Premier League season.
Marcelo Bielsa may have turned Leeds’ fortunes around in his three seasons at the club, but one thing that he has never seemingly managed to sort out during his tenure is Leeds’ set pieces.
There were times this season when the team looked extremely vulnerable from set pieces. During December whenever the team conceded a corner or free kick, Sky Sports would present the graphic showing how poor Leeds are from set pieces, and it felt like a goal was coming every time. Whilst I couldn’t find the graphic in question, the one below shows just how bad Leeds were at defending set pieces.
Leeds conceded 15 goals from set pieces which was the highest total in the Premier League during the 2020/21 season, 5.5 goals above the league average, so if you’re a “the only stat that matters is goals” person, then Leeds aren’t looking the best so far.
On the offensive side of the ball the picture looks better, as can be seen in the chart below:
Leeds ranked joint 7th for set-piece goals scored last year with 11 goals scored, which was 1.5 goals above the league average. This was only one behind Chelsea who many would consider one of the best teams from set pieces last season.
What can’t be concluded just from looking at set-piece goals scored and conceded is if this fairly represents a team’s ability at set pieces or if luck played a part. In this piece, I’ll look into the data in more detail to see if Leeds are as bad at defending as their goals conceded suggests and if the team has shown any improvement throughout the season.
The underlying numbers
Focusing on Leeds’ attacking Set Piece Expected Goals first, the team accumulated 10.26 Expected Goals meaning they slightly over-performed last season. Against the other teams in the league, Leeds ranked eighth in the league for Set Piece Expected Goals.
On the defensive side, the underlying numbers back up that Leeds were poor at defending set pieces last season.
15.2 Expected Goals conceded from set pieces was the joint worst in the Premier League and meant Leeds actually over-performed when defending set pieces.
The graphic below from StatsBomb’s Manchester United Season Preview shows the quality of chances each team in the league created and conceded from Set Pieces last season.
In the graphic, it can be seen that the quality of chances Leeds conceded from set pieces was the second-worst in the league which will have contributed to Leeds ranking worst in the league for expected goals conceded. Leeds created 0.0067 xG/Set Piece which was near the average for the league.
If we now focus on the individual match data, the Expected Goals and Expected Goals Conceded data for each match can be seen in the graphics below. An eight-game moving average is also shown on both charts.
Looking at the above graphics, Leeds’ performance at set pieces both offensively and defensively can be split into four sections. Set-piece performance was at its best in the first ten games of the season (up to Everton away) as both the xG and xGC moving average were at or near their best values at this point.
From game eleven (Chelsea away) to game twenty-one (Everton at home), the team’s set-piece performance declined, with both the xG and xGC moving average being at or near their worst point at game twenty-one. The Chelsea away game was the most Expected Goals Leeds conceded from set pieces during one game last season.
After this point until game twenty-eight (Chelsea at home), set-piece performance improved. The xG moving average returned to near its best value. However, the xGC was still nearly double its best point of the season. Leeds produced their highest xG from set pieces during the Crystal Palace home game. From game twenty-nine (Fulham away) till the end of the season, the xG trend stayed fairly consistent whilst the xGC trend began to decline once again.
Why does the data look like this?
Before I try to find an explanation for the data and trends shown it’s worth a brief explanation of some of the principles Leeds use at set pieces.
Every player is man-marking an opposition player as Leeds do during open play, with Bamford the only player without a man-marking task meaning he is free to attack the ball. This also allows him to be free to counter quickly. One of the wingers hugs the touchline near the corner flag, effectively man-marking the corner taker but also allowing them to counter quickly. Man marking at set pieces for Leeds is, therefore, more about the attacking opportunities it brings rather than because it makes sense for them defensively.
Due to the man-marking system, if a player loses his man this can often leave the opposition player with a relatively easy chance to score, which explains why Leeds performed so badly for xGC/Set Piece.
The man-marking system is also reliant on a defensive player’s ability to win aerial duels against the player they are marking which unfortunately for Leeds is an area where they struggle. The first reason for this is that the squad is not blessed with many tall players. Outside of the centre backs, it is only Klich, Ayling and Dallas who are taller than six foot (and in the case of the latter two players it is only just). The second reason is due to the ability of the Leeds defenders relative to the opposition players who scored against them last season.
SmarterScout uses a rating system to evaluate players’ skills in aerial duels from dead balls. The system is similar to the ranking system for chess players in that when two players enter a one-on-one competition for the ball one player will win. A higher-skilled player beating a lower-skilled player is expected whereas the reverse is not. In the first situation, the higher skilled player’s rating will rise less than it would fall in the second situation. You can read more about the rating system here.
A comparison of the ratings of Leeds’ defenders and the opposition players to score from set pieces against Leeds can be seen below.
Looking at the above ratings, it is clear that Leeds’ struggles at set pieces are partially down to the poor ability of the current squad at set pieces. But they are also down to conceding to some of the best players in the league from set pieces. Therefore, when these players are matched up due to the man-marking system, Leeds are going to struggle and this is reflected in the number of set-piece goals conceded.
This was part of the reason Leeds’ Expected Goals Conceded increased from the eleventh match onwards. Luke Ayling played a run of games at centre back from this point, and this coincided with Leeds facing Chelsea and West Ham who (according to xG) were the two best teams in the league at offensive set pieces, due to the threat of players like Kurt Zouma and Tomas Soucek. Once Ayling returned to right-back and the run of fixtures eased from a set-piece perspective Leeds were not as poor from set pieces. This coincided with Meslier seeming to come and claim corners more frequently which helped ease the opposition threat.
On the Offence
Offensively, Leeds do not seem to do anything complicated at set pieces. Set-piece delivery is therefore crucial to the team creating chances.
Raphinha and Kalvin Phillips were Leeds’ two main set-piece takers last season, with Raphinha performing better as he created 26 shot-creating actions from dead balls last season (according to FBref.com) compared to 19 from Phillips. Leeds’ improved xG performance from match twenty-one is also the point at which Raphinha begins to take the majority of Leeds’ set pieces, as can be seen in the graphic below.
The combination of improved delivery from set pieces was also helped by the easier fixtures and the return of a recognised centre back leading to increased threat from set pieces for Leeds.
Due to being so reliant on good delivery, this would explain why Leeds are only average performers in terms of xG from set pieces but also xG/Set Piece.
What can we expect this season?
Whilst it seemed that Leeds improved defensively from set pieces in the second half of the season, the underlying numbers disagree, meaning Leeds just got luckier if anything.
The only improvement that has been made over the summer is Junior Firpo who will help the height issue. However, Firpo is not noticeably good at aerial duels meaning that this is unlikely to lead to much improvement. The man-marking system means that Leeds will continue to concede big chances so unless any other changes are made defensively I would expect Leeds to continue to struggle.
On the attacking side, I would expect Leeds to perform similarly to how they did last season. A full season of Raphinha being the primary set-piece taker may lead to a small increase in xG created from set pieces. However, without any other changes, I don’t think the quality of chances Leeds create will increase.
You can follow Tom Alderson on Twitter @TomAlderson97 for more content.
If you enjoyed this article 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.
