Skip to main content
Football Statsbomb Performance Analysis

Digging Deeper Into Ball Carrying

9 min Read

In this article, we look how Hudl Statsbomb data can be leveraged to identify valuable traits in ball carrying players.

If I wanted to explore a player’s passing in detail, there are several models that could help guide my analysis. I could use xPass to find out the likelihood of a given pass being completed, or pass clustering to group similar passes together and identify patterns. Within a few lines of code, I could understand how likely a passer is to take risks, the quality of their passing relative to expectation, the types of passes they typically make, and common passing relationships.

Compared to passes, carries are more unwieldy, a moment of inspiration and improvisation that can cut through defenses or lead to dead ends. I wanted to take some time to explore how the Hudl Statsbomb dataset could add context to ball carrying and help identify valuable traits in ball carrying players. This is by no means definitive, but if I wanted to know more about how a player carried the ball, here are some options for analysis I might take.

All carries in this analysis have a minimum length of five yards, unless otherwise stated, and the dataset used is from the Premier League in 2024/25, up to matchweek 29.

Sort descending and call it a day

Luckily, there is a lot in the dataset that can be used to inform our understanding of ball carrying. Statsbomb’s On-Ball Value (OBV) model - which assigns a value to each action based on how much it increases or decreases their team’s impact of scoring and conceding - is a useful shortcut for quickly identifying strong carriers. 

Take a look at the top 20 Premier League players for Dribble & Carry OBV and you’ll see a list of some of the most creative (and occasionally frustrating!) players in the league. Sávio stands out for his particularly high output, but there isn’t much separating the rest of the pack.

This leaderboard is a natural starting point for identifying effective ball carriers. If we wanted to understand in more detail how these players use carries in their game, one option would be to look at the actions that they took upon completing a carry. This is a logical first step as carries are an intermediary action taken between receiving and releasing the ball. We can take advantage of Statsbomb’s xG and OBV models to identify the quality and value of subsequent shots and passes respectively.

Because we’re looking at the top carriers in the league, there is not a large difference in each player’s output post-carry, but there are some insights here. We notice that some carriers are more likely to create post-carry value via shooting, while others create more value from passing. This analysis also allows us to identify some players lower down on the leaderboard who stand out for creating a lot of value from their actions post-carry, such as Harvey Barnes and Heung-Min Son

The side effect of this analysis is identifying a subset of carries that preceded a shot, which can be considered in video analysis or plotted visually. In the example below, we can see Antoine Semenyo’s carries that preceded shots. Semenyo is a shooting threat from either wing, typically drifting inward towards the shoulder of the box. He also occasionally shoots from short carries originating in zone 14. 

Counter-Attacking Carriers

We can extend the idea of looking at the event following a carry to consider carries that formed part of a successful sequence of play. As it would be complex to truly understand how relevant a given carry was to a possession chain, particularly in a long sequence, let’s consider a particular type of possession sequence where we might be able to value a carrier’s contribution with greater confidence: counter-attacks.

The majority of counter attacks centre on a player driving forward with a carry from deep, while teammates charge forward to provide passing options and opponents scramble to rectify their defensive shape. As counter attacks are quick and direct, a carry that forms part of a counter attacking sequence is likely key to the success or failure of that pattern of play. 

A major caveat here is that these numbers will be influenced by the team’s play style: players capable of carrying in a counter attacking context will register lower numbers if they are part of a team that does not typically counter. Nonetheless, I thought it could be interesting to see how these numbers looked for players within a team:  Liverpool, the team who generate the most counter attacking shots in the league. Mohamed Salah, who narrowly missed out on the top 20 for D&C OBV, leads here. 

Interpreting Space

In exploring options for this analysis, I watched a lot of videos of carries. It became apparent that successful carries require good spatial awareness. Wasted carries tend to be ones where the player runs into congestion or holds onto the ball too long, while capable ball carriers release the ball at the right moment, while still in a pocket of space.

Typically, a carrier sacrifices space to achieve progression, but by maintaining at least some distance from opposition defenders, the carrier of the ball can pick out a pass or take a shot without being dispossessed. 

Let’s take a pocket of space to be a carry end location of at least 2 metres from the nearest defender and focus on the final 3rd, where space is hard to come by and also most valuable. We can plot this alongside the median space a player is in at the start of their carries into the final 3rd to further contextualise how players are using space. 

Within our cohort of strong ball carriers, some players are more regularly carrying into pockets of space than others. Broadly speaking, players who start in more space will more regularly maintain that space as they progress the ball. Sávio and Jeremy Doku, the league’s top carriers per OBV, stand out for most regularly carrying the ball into pockets of space in the final 3rd, despite starting from less space.

Plotting these carries into a pocket of space visually reveals certain areas where a given player is more likely to find space through a carry. Sávio, for example, is typically already in the final third when he carries into a pocket of space. He uses both wings, but typically targets the right shoulder of the box. There are also a handful of instances where Sávio carries in towards the centre, and some indication that he will occasionally drop backwards into space.

Including the space a player is in at the start of a carry introduces another angle for analysis. As well as understanding how players conserve space at the end point of a carry, we can explore how the amount of space a player is in at the point of ball receipt influences their decision to attempt a carry. In turn, we can identify players that are capable of creating danger through carrying even when in tight space. 

To begin to explore this, we can start by applying the same techniques used to assess carries in general to a subset of carries that started in tight space. Highlighted below is the OBV for the league’s top carriers when in tight space (less than 3 yards from the nearest defender). Immediately apparent are significant differences in how much value a player adds when in tight space, with Saka, Doku, and Díaz standing out.

Perhaps high OBV players that do not generate as much OBV when in tight space can be neutralised through targeted pressing, while players creating OBV from tight space could be considered particularly dangerous. We might also seek to understand how a player responds to tight space in ways that alleviate pressure and ensure possession retention. There will be some situations where, even for a player adept at working in tight space, a safe backwards pass is the smartest option. 

One on One

We can also consider how to define space in the data to identify specific scenarios. We typically define space by the distance to the nearest defender, but we can be more specific by looking at the subset of tight positions where the nearest defender is ahead of the player, thus cutting off the forward route they would usually take. For situations where the player chooses to take on the opponent directly, these situations will typically show up as dribbles in the dataset.

Amongst these players, Jeremy Doku stands out for being particularly inclined to attempt dribbles and completing them at an above average rate. There are some indications that players more likely to win dribbles attempt dribbles at a higher rate, with some exceptions. Callum Hudson-Odoi, for example, has a strong success rate when attempting to dribble, but is less likely to attempt dribbles than most of the other players in this group. 

We might then seek to understand the alternative actions a player might take in these situations when choosing not to dribble. Plotted below are Callum Hudson-Odoi’s carries attempted when facing a close defender in front of him. Hudson-Odoi appears to have a tendency to retreat and pick out a pass towards the centre of the pitch. He is seldom dispossessed or misplaces passes from these situations, all of which is indicative of a tendency to play it safe when facing a defender head on.

Given that this is a cluttered visual where the analysis relies on picking out specific examples, I feel there is more work to be done. We would need to develop metrics that capture the various behavioural tendencies from these situations, allowing us to compare players statistically. One option could be to look at a player’s dispossession ratio when facing a defender in tight space. Another could be to look at the xPass of passes attempted from these situations. Is the player typically attempting risky or safe passes post carry?  Employing OBV both for the carries themselves and the actions that follow them could help model this.

Conclusion

Some combination of the approaches discussed above should help us better understand how a given ball carrier operates and the impact that their carries have on the success of their team. 

In working on this piece, two options for this type of exploratory analysis emerged. One is to identify a trait in match footage and then explore how the data available could be used to define it (see carries into pockets of space). Another is to consider the data models available and speculatively run some analysis to see if anything interesting is revealed (see OBV of next action after a carry). Both of these approaches are a worthwhile way to get started looking into something new.

Do you work for a club, league or federation? Want to get a hands-on feel for the value of Hudl Statsbomb data? Get in touch today to request a free trial.