Packing in football: Leicester City
To start, what is packing in football? Well, firstly, as an aside, I’m not going to proclaim to be an expert on this topic. I’m a late explorer of the packing concept, which began after a talk with Alan Morrison (a Scottish journalist/analyst covering Celtic FC). Since our conversation, I’ve enjoyed transferring this idea onto Leicester City for a handful (seven) of games this season.
The idea of packing was first invented by former professional footballer, Stefan Reinartz (a German international), to develop a detailed understanding of passing and dribbling, by adding contextual values — described as the packing score, to an action due to how many opposition player(s) are bypassed en route to a destination, either the end of a dribbling action or the recipient of a pass.
Its origins date back to Euro 2016. German TV stations displayed their normal game-to-game statistics (possession, goals, shots, etc.), with an extra add-on in the form of packing score as they partnered up with Reinartz’s then-newly founded, Impect GmbH. A data provider solely focused on revolutionising the idea of passing and dribbling through packing.
In short, the concept is fairly simple but provides a lot more insight than other metrics on the subject. For example, progressive passes or passes into the final third both require location, or distance-based proxies, to be met, but offer no information on how these passes interacted with the opposition. Whereas, packing does. If a single player is bypassed with a pass, or dribble, then the packing score for that action is 1 × a position-based modifier.
The position modifier works to balance out passes that break the defensive line, versus “easier” passes in midfield. If a defender is bypassed with an action, that is a score of 3, while a midfielder is a score of 2, and a forward a score of 1 — when multiple players are bypassed with a single action, the packing score is added together. This becomes slightly fluid when looking into formations, structures, and how teams differ between their in-possession and out-of-possession shapes.
Since its infancy nearly eight years ago, packing has struggled to puncture the mainstream data sets, still being fairly exclusive to Impect GmbH. A major reason for that is the conditions of how the data is collected. For instance, to enable an examination of Leicester City through this lens, I’ve had to rewatch games and code them manually, entirely focused on passes/dribbles and the opposition structure, as it’s imperative to the data that you react to the opposition player(s), not simply the pass completion or the distance of a pass.
Later in this article, I will provide a short explanation of how the data was collected, as I think that’s important to understand the results I’ve found, whilst also highlighting a potential shortcoming — it’s a manual task, so different collectors of the data will perhaps score passes in different ways. For my results, there’s consistency in the way I coded the matches, but cross-comparing with other sources might unlock interesting and unfair debates.
How I’ve interpreted packing
There’s a brief synopsis of the idea of packing above, but let’s talk through it. It’s an extension of the dribble and passing metrics, to develop an understanding of how penetrative a single player or team's actions have been.
Above I’ve added a simple diagram illustrating two different passes and their associated packing score. As you can see, pass #1 (on the left side) has a fairly low packing score of 3. It’s bypassed the pressing forward (1), and the recipient has received possession behind one of the midfielders (2), totaling a score of 3.
Pass #2 is a higher-value example, with only two extra players bypassed. The ball player has removed a line of three midfielders (3 x 2 = 6) and also played beyond a forward (1), for a packing score of 7. Neither of these examples are pulled from a Leicester City game, they’re designed to help articulate how the metric works.
However, when watching a full 90-minute game of football, a large majority of the passes (and the occasional dribble) don’t move the ball beyond a player or metaphorical line. In packing score terms, these passes aren’t counted, but by accessing the raw data from the games, I could cross-reference the percentage of passes that reach the packing definition, and by default interact with the opposition structure.
Again, this will be another reason my results are fairly isolated. I’ve chosen to interpret the data in a multitude of ways, perhaps others would make different choices with how to showcase the information. As an analyst, I enjoyed the freedom this presented. I could create random thresholds, or potentially sub-metrics, that attempt to illustrate how an individual or team passes/dribbles in possession.
Collection Process
To start this data set, I had to find unedited footage of Leicester City’s games this season. Which, without an online library (which every other club in the Championship has, either free of charge or for a small monthly fee) has restricted the total number of available games I’ve been able to analyse.
In total, I’ve used seven games (vs. Sheffield Wednesday (A), Rotherham United (H), Leeds United (H), Middlesbrough (A), Ipswich Town (A), Cardiff City (A), and Birmingham City (A)). This is partly due to the time it takes to conduct the analysis (I estimate each game takes an average of 3–4 hours). The other reason is that this piece is aimed at being an introductory analysis of the subject, so I didn’t want to watch all of the Foxes' available games this season, as that would’ve only prolonged the release of the article.
I won’t bore you with the details of how I map the data from the game perspective, it’s a lot of freeze-framing passes, and calculating the associated number of opposition players bypassed. To ensure consistency, there’s a lot of rewinding and double analysing, but the data collected might differ from others watching the same footage — I would argue that my data is accurate for all the prerequisites I set for collection. Anyway, that’s enough pre-rambling, let’s look at the data.
Findings
In a metric designed to highlight the number of opposition players a passer and its recipient combine for, it’s unsurprising that the stat leaders are all defenders. In most of the sequences where the defenders hold possession, the opposition's entire block is goal-side of the ball, so theoretically they have ten outfield players to bypass with a pass or dribble.
Even with the position modifying scaling, the line spacing of the midfield/forward line means that Leicester City’s defenders have more freedom to break the lines. To further prove this, I decided to calculate the percentage of a player's passes that are packed, which is displayed below.
Conor Coady, who in the seven-game sample size has played 198 minutes, leads the Foxes for the highest percentage (%) of passes that are packed. He is closely followed by Wout Faes (634 minutes), with both players averaging above 28% of their passes bypassing an opposition player. Of the top five, the majority of the players function as CBs in Enzo Maresca’s 3–2–4–1 formation, but the exception to that rule is Stephy Mavididi.
As a comparison, Abdul Fatawu plays a similar role to Mavididi but from the right side of midfield. With comparable minutes in the sample (Mavididi: 661 to Fatawu: 550), both players attempt slightly more than 30 passes per 90 (p/90). Despite this, Mavididi completed 8.31 packed passes p/90, or for the above visual, 24.3% of his total completed passes. That’s vastly different from Fatawu’s output of 4.42 packed passes p/90 or 13.92% of his passes.
However, it’s simply frequency. It’s very impressive how functional Mavididi’s packed passes are, especially when considering the system deployed for the right-side creation involves heavy rotations into the right half-space from Wilfred Ndidi as an underlapping runner. It’s mirrored by Kiernan Dewsbury-Hall, but the eye test suggests that the frequency tends to skew to the right, but Fatawu’s ability to find packed passes isn’t as effective as his wing partner. But, which players are more productive with their packed passes?
It’s the same name leading the charts for Leicester City, Conor Coady. For each pass Coady completed (204), he averaged a staggering 1.56 packing score p/ pass. Not to keep returning to the winger dynamic, but you can see Mavididi’s 1.2 packing score p/ pass is better than Fatawu’s 0.76 packing score p/ pass. It’s not just frequency, but also the output for the former Montpellier left winger.
While this chart does a good job of interpreting both the players that make a lot of passes (defenders) and consequently evaluating how effective they are at finding teammates beyond an opposition player, this chart still favours deeper possession players. As an alternate view, I wanted to change the focus from being total completed passes, and instead look at when a player makes a packed pass and how penetrative they are.
There’s finally a different leader, this time it’s the fairly inactive Fatawu (13.92% of his total passes or an average of 4.42 p/ game) is extremely penetrative when executing a packed pass — averaging a 5.44 packing score, with Coady (5.39) very close in second place.
To break down the numbers, Fatawu beats the equivalent of 5.44 forwards, 2.72 midfielders, or 1.8 defenders with each of his packed passes. Considering that his reception of the ball tends to be in the second-to-third phase of a possession sequence, Fatawu’s passes don’t tend to interact with the opposition forwards. In fact, of his 27 packed passes, only two of them bypassed an opposition forward.
It’s his split of defenders and midfielders bypassed that I find very intriguing. Most players, irrespective of positioning, tend to bypass the midfield line with higher levels of frequency than the defensive line. Fatawu’s passes have bypassed 32 midfielders and nearly an identical 27 defenders. The position modifier places opposition defenders as significantly more difficult to pass beyond than forwards or midfielders, yet Fatawu completes a large portion of his packed passes beyond defenders. Again, he’s a fairly inactive-packed passer, so that perhaps explains that.
Aside from Fatawu, the top five still offer a sense of variety. Two CBs, Conor Coady and Jannik Vestergaard, and Leicester City’s double pivot #6 partnership, Harry Winks and Ricardo Pereira. If you understand Maresca’s building system, which involves holding possession centrally and poking for progression, it’s pleasing to see that the data seeds these exact players in this fashion. However, it’s still deep possession players that reign supreme, so if we flip the narrative how does the data stack up?
If the players in depth tend to lead the packed passes charts, then the recipients can also be mapped, giving a chance for the forward players to be valued in how they make space to receive beyond opposition players. Two players lead the pack on the above chart, Jamie Vardy (7.3) and Patson Daka (7.21), two of Maresca’s #9’s.
I might do a longer article surrounding the striker position, as I think that’s an area of interest/focus for fans, particularly with a supposed lack of output from those playing major minutes in that position. For what it’s worth, in my opinion, Maresca wants a forward capable of playing both the “to feet” and “in-behind runner” roles or as I’ve started summarising, a #9.5.
To definition, Vardy resembles a stereotypical #9, while Kelechi Iheanacho borders on being a #10. I think Daka, who receives just shy of 20 packed passes p/90, has been the best blend of the dual in-possession roles, but maybe Tom Cannon is the long-term solution. Either way, Cannon isn’t represented on the chart, as his 31 minutes didn’t meet the 1 >/ 90s threshold over the seven-game sample sizes. Stylistically, his 7.2 packing score per pass received partners him with Daka and Vardy — last-line receivers.
Whereas, Iheanacho’s (5.54) pass reception numbers bear resemblances to the high #8’s in the side. This data provides a succinct way of highlighting the advanced nature of reception, but also the tendency of a player (i.e. Iheanacho vs the other forwards). This data set lends itself to player analysis, particularly within the squad. However, I did track the overall team statistics and collected both the individual/team for the opposition also.
As an average, Leicester City have a packed pass percentage of 22.01%, which in six of the games placed them with a lower percentage than their opponent — only the game vs. Cardiff City (A) saw the opposition (20.79%) have a lower packed pass percentage than the Foxes, both in terms of the seven-game average, but also the individual fixture. The range was at its lowest 19.87% vs. Sheffield Wednesday (A), and a high of 26.42% vs. Birmingham City (A). It’s difficult to suggest whether these numbers subvert the norm, as I don’t have the quantity size to make such suggestions.
Generally, Leicester City makes less packed passes as a percentage than their opponents, with the added context that Maresca’s side has seen the majority of possession, and tends to complete a lot more passes. Even though the data is presented as a percentage, fewer passes tend to increase the efficiency of packed passes, as directness correlates to higher packed numbers and teams attempting fewer passes tend to play more directly.
This is, much, much longer than I intended, but hopefully explains the idea of packing and starts to explore how Leicester City’s players performed in this metric. I will continue to collect and use packing in my work, so this article might be linked/referenced quite frequently to help illustrate my usage of the metric to those reading it at a later date.
