Football Analytics Redefined: Leveraging the xPass Metric for Insights

Using the StatsBomb’s xPass metric in football analytics to reveal how the difficulty of passes attempted during different game situations reflects tactical decisions and mental states.

Aug 09, 2023

Introduction

The article’s main objective is the exploration of a novel football analytics concept based on StatsBomb’s xPass metric. The author reflects on the inception of this idea when StatsBomb introduced the xPass metric (Pass Success Probability), prompting their interest in uncovering intriguing insights from it.

Can the difficulty of passes (as indicated by xPass) made by players or teams during various game states provide insights into their playing style and mindset?

The primary questions the study seeks to address include whether the nature of passes attempted by players or teams during different game situations reflects their strategic approach and mentality. This raises inquiries into how successful teams manage their lead, whether specific players exhibit distinct behaviors under different match circumstances, and how teams respond and adhere to their philosophies when faced with adversity.

What is an xPass?

The xPass metric is introduced as a quantification of the estimated probability that an attempted pass will be successfully completed. This estimation is based on several factors, including the pass’s location on the pitch, the contextual conditions under which the pass is executed, and the intended target location of the pass.

In its basic interpretation, the xPass metric serves as a measure of the level of difficulty associated with a pass. Passes with higher xPass values are expected to be completed with a high success rate, representing safe and low-risk passes. An example of this would be a routine pass between two center-backs in an unpressured situation, which would yield a high xPass value.

On the other hand, passes with lower xPass values are anticipated to have a lower success rate, signifying riskier and more ambitious passes. A scenario could involve a full-back attempting a pass into the channels from their own half while under pressure from an opponent, leading to a lower xPass value.

In broader terms, the metric’s interpretation suggests that a higher frequency of high xPass passes reflects a measured possession strategy characterized by safer passes. Conversely, a higher prevalence of passes with lower xPass values indicates a more direct and potentially risk-taking style of play.

What can we do with xPass?

Next, the author delves into the potential applications of the xPass metric in understanding team playing styles, particularly in different game states. The author aims to explore whether xPass can serve as a proxy for identifying team styles and whether these styles change based on the game state (winning, losing, drawing). The author employs visualizations and data analysis to address these questions.

The visual representation provides a comparative view of teams’ average xPass values across different game states: winning, losing, and drawing. The percentages of passes played in each game state are also presented, offering context for teams that frequently experience specific game states. The teams are ordered based on their average xPass value during drawing game states, and league averages per game state are indicated by dashed lines. The analysis is conducted both holistically and considering teams under different managers, with a minimum threshold of 10 games.

The analysis yields several key insights:

1. Drawing Game State Reflects Intended Style: The drawing game state emerges as the closest representation of a team’s intended style of play. This equilibrium between attacking for a win and defending to avoid conceding aligns with most teams’ typical circumstances during matches.

2. Opponent Influence on Pass Types: The passes executed by a team might not solely reflect their preferred style. Opponent strategies and tactics play a role in determining pass types. For instance, a winning team might intentionally allow the opponent possession to conserve their lead, influencing the passes made.

3. Flexibility of Team Styles: Teams may adapt their playing style based on the opponent’s ranking or whether they are playing home or away. This dynamic nature of team styles should be acknowledged when analyzing aggregated data.

Considering these observations, the author makes several assumptions:

1. Teams with higher positions in the visualizations tend to adopt a possession-based style of play.

2. Teams positioned lower in the visuals tend to favor a more direct playing style.

3. Markers leftward of a team’s xPass during drawing suggest a more expansive or direct approach compared to their drawing style.

4. Markers rightward of a team’s xPass during drawing indicate a more conservative or intricate style in that game state.

This section demonstrates how the xPass metric, coupled with thoughtful analysis, can provide insights into team playing styles and how they vary across different game states. The underlying nuances of tactics, opponent behavior, and contextual factors shape these findings.

What does xPass tell us?

In this section the author presents a comprehensive analysis of the insights gleaned from applying the xPass metric to soccer match data. The author examines the xPass values of teams across different game states, specifically drawing attention to higher xPass values when drawing and the variations in playing style when winning or losing. The findings reveal valuable information about team strategies, especially for weaker teams and those under different managers.

One prominent observation is that teams with higher average xPass values when drawing tend to maintain a consistent style of play across all game states, regardless of winning or losing. Conversely, teams with lower xPass values when drawing exhibit greater stylistic variations based on game state, resulting in a wider dispersion of average xPass totals within this group.

The analysis uncovers that weaker teams, when leading, tend to adopt a more direct and riskier style of play. This strategic shift is intuitive, reflecting a priority to safeguard their lead by moving the ball away from their own goal, even if it increases the likelihood of losing possession. This tactical choice emphasizes territorial gains over ball control. However, the author points out that playing a direct style entails nuances; different teams execute this approach with varying levels of effectiveness.

Comparing clubs that changed managers mid-season, the research uncovers intriguing insights. For example, the comparison between Frank Lampard and Sean Dyche’s Everton teams highlights a considerable stylistic difference, especially in changes to xPass values when winning or losing. The analysis also underscores the shift in Everton’s underlying metrics, showcasing how Dyche’s influence led to both tactical changes and improved performance.

Villarreal’s case demonstrates managerial continuity in terms of xPass values despite a change in leadership, resulting in a commendable 5th place finish in La Liga. This underscores the hierarchy’s effective identification of a suitable successor to maintain the team’s strategic consistency.

An intriguing outlier is Raffaele Palladino’s Monza side in Serie A. This team showcases a possession-heavy approach when drawing and, interestingly, intensifies this style even when trailing. This patience and persistence in maintaining their playing philosophy paid off, as Monza secured an impressive 11th place finish in their inaugural Serie A season.

Overall, this section highlights the depth of insights that can be extracted from xPass data analysis. It showcases the nuanced dynamics of teams’ playing styles in various game states, shedding light on how strategies evolve, adapt, and ultimately contribute to performance outcomes.

Does xPass matter?

What is the significance of xPass data? What are the implications for teams’ playing styles and performances? Here it is underscored how stronger teams tend to exhibit certain patterns in their xPass values across different game states, reflecting their possession-oriented strategies and player quality. Notably, even teams positioned lower in the xPass rankings tend to demonstrate consistent playing styles across game states, indicating a level of tactical stability.

The observation that stronger teams maintain a relatively consistent xPass value regardless of game state led the author to explore this phenomenon further. To investigate this theory, the author introduced a metric called the “Divergence Score,” which quantifies the variance between a team’s average xPass values when drawing and their values when winning or losing. This Divergence Score serves as an indicator of how teams adjust their playing styles based on game situations.

By plotting the Divergence Score against a team’s season xG Difference, a notable correlation emerges. The data suggests that a higher xG Difference, indicating a team’s superior performance, is associated with a lower Divergence Score. This finding implies that successful teams tend to maintain their playing style consistently across different game states, regardless of their lead or deficit. The graph visually demonstrates this correlation, with a concentration of high-performing teams in the top-right quadrant.

This analysis concludes by proposing the idea that success in soccer involves not only enduring hits and progressing forward but also maintaining a consistent tactical approach even when facing challenges. The data insights presented emphasize the interplay between strategic adaptability and consistent playstyle, shedding light on the complex dynamics that contribute to teams’ performance outcomes.

How can xPass help us?

In this last section, the author tries to answer why any of this matters and what actionable insights can be generated.

Manager assessment

The text then examines the practical implications and applications of xPass data analysis, particularly in the context of evaluating managerial effectiveness and identifying potential candidates. The author discusses the potential utility of the Divergence Score metric in assessing managerial performance and drawing insights for decision-making processes.

The author proposes that analyzing the Divergence Score could provide valuable insights into managerial effectiveness. A low Divergence Score, indicative of a consistent playing style across different game states, could suggest several positive attributes:

1. Effective Communication and Implementation: A manager with a low Divergence Score likely has effectively communicated and instilled their tactical intentions within the team. Players understand and execute the gameplan consistently.

2. Trust and Cohesion: The low Divergence Score may also reflect a strong level of trust and cohesion within the team. Players adhere to the manager’s strategy even in challenging situations, demonstrating their belief in the approach.

The author acknowledges the counterargument that such consistency might indicate a lack of adaptability or “plan B.” This perspective has been associated with managers like Mikel Arteta and Graham Potter. However, the author suggests that if the consistent approach is coupled with positive on-pitch results (measured by metrics like xG, OBV, or points won), it could indicate successful managerial strategies.

Furthermore, the author envisions this analysis as a potential tool for identifying emerging managerial talents. By conducting similar analyses on leagues beyond the top five and considering different metrics, clubs could potentially identify managers who demonstrate tactical consistency alongside strong results.

It’s important to note that a high Divergence Score does not necessarily indicate poor managerial performance. A manager-team combination with higher divergence might have intentional strategies for different game states, emphasizing adaptability and tactical variation. The critical point is that Divergence Score analysis provides valuable insights that, when interpreted in conjunction with other metrics, can contribute to a more comprehensive understanding of managerial capabilities and team dynamics.

Performance analysis

Here the focus shifts towards the practical application of the derived insights in a club setting, particularly emphasizing the significance of context-rich information for understanding team performance and making informed decisions.

For clubs fortunate enough to have access to the additional context provided by their first-team manager and coaching staff, the presented data becomes a powerful tool for analyzing team performance. The discussion zooms in on Swansea’s playing style, which is distinct and characterized by a low Divergence Score, reflecting consistency across different game states. However, an anomaly surfaces in a match against Luton, where Swansea experienced an unexpected 1–0 loss and had a mere 4% chance of winning according to xG.

The analysis delves into potential reasons for this unexpected outcome. It raises questions about Swansea’s preparedness for Luton’s physical style of play and whether there was an intentional alteration in tactics that misfired. The idea of assessing these aspects using insights from data analysis emerges as a pathway toward rectifying similar mistakes in the future.

Additionally, the section highlights the reverse application of this analysis. If Swansea is the next opponent, the gathered information and insights from their match against Luton could offer a strategic advantage. By meticulously examining videos and diving deeper into the data, clubs can pinpoint factors that disrupted Swansea’s typically strong alignment with their playing style. This, in turn, can guide the opposing team in formulating tactical approaches and exploiting vulnerabilities.

Player analysis

This section delves into the potential of the discussed data analysis approach to provide insights into individual players’ behaviors, particularly in the contexts of player recruitment and analyzing opponents’ strategies. The focus is on understanding how analyzing xPass and Divergence Scores can shed light on players’ playing styles and tendencies.

The author poses two significant questions that this analysis can address:

1. Recruitment Insights: Can the analysis of individual players’ passing patterns based on xPass and Divergence Scores provide valuable insights for player recruitment? Specifically, can these metrics help clubs understand how players adapt their passing strategies based on game states (winning, drawing, or losing)? The author employs the example of two Centre Backs from Sunderland, Danny Batth and Dan Ballard, to illustrate this concept.

— Danny Batth: Batth’s passing style differs depending on the game state. He tends to play similar passes when the team is drawing or losing, but he adopts a more direct approach when the team is winning.

— Dan Ballard: Ballard, on the other hand, maintains a more consistent passing style when the team is winning compared to when they are drawing. Additionally, he displays more patience and thoughtfulness in his passing when the team is losing.

These distinct patterns in passing behavior offer potential insights into players’ adaptability and strategic decisions based on the match situation. This data could be useful for clubs when assessing potential recruits. By understanding how players adjust their style depending on the game state, clubs can evaluate whether a player’s tendencies align with their team’s tactical approach.

2. Opponent Analysis and Strategy: The analysis of players’ passing patterns also has implications for analyzing opponents and devising match strategies. The author raises the question of how understanding individual players’ playing styles can inform a team’s approach in pressing an opponent’s backline.

For instance, if an upcoming match involves playing against Sunderland, the analysis of the distinctive passing behaviors of players like Batth and Ballard could influence how the opposing team chooses to press their backline. Recognizing that Batth tends to go more direct when winning and Ballard remains patient in his passing when losing can inform defensive strategies to disrupt their preferred patterns.

Furthermore, the author explores the concept of tailoring game strategies based on the player occupying a specific role. Different players in the same position might exhibit varying passing tendencies. The example cited involves considering whether Edouard Michut, Corry Evans, or Dan Neil is playing in the CDM role. Their distinct playing styles and passing behaviors could lead to different tactical considerations during the match.

Recruitment

Finally, it is explored the potential utility of the xPass and Divergence Score analysis in the player recruitment process, specifically when identifying players for certain roles and playing styles. This section highlights the potential of the analytical approach to inform recruitment decisions and focuses on the role of a Central Defensive Midfielder (CDM) as an illustrative example.

The primary question posed here is whether the insights derived from analyzing xPass and Divergence Scores can serve as valuable factors in the recruitment process. In particular, the example revolves around the search for a CDM who excels in slowing down the game in the midfield. This role demands a player who can maintain composure under pressure, retain possession effectively, and contribute to the recycling of the ball while enabling more creative teammates to advance play.

The discussion then shifts to the practical application of the analytical insights. Using Edouard Michut as an example, the paper suggests that his consistently high xPass values across different game states could indicate his suitability for the specified role of a CDM who can control the tempo of the game and sustain possession in the midfield. This observation suggests that Michut’s playing style aligns with the requirements of the role being sought.

The broader implication of this scenario is that xPass analysis can provide a quantitative basis for assessing a player’s potential fit for a particular role. In this case, the elevated xPass values across various game states signal Michut’s inclination toward measured and composed passing, qualities that align with the role of a possession-retaining CDM. The paper thus introduces the notion that xPass analysis can serve as a tool for enhancing the effectiveness of the player recruitment process by providing valuable data-driven insights into the playing style and suitability of potential recruits for specific tactical roles within a team.

Conclusion

In conclusion, the presented analysis offers a preliminary exploration into the potential insights that can be gained from associating xPass values with different game states. While acknowledging the need for further in-depth investigation, the author has laid the foundation for a thought-provoking discussion and encourages the readers to contribute their suggestions and insights for refining the methodology.

The significance of linking xPass values to game states is underscored by its potential to unveil valuable information about team and player styles. This information holds relevance across various domains including league analysis, team dynamics assessment, managerial evaluation, and player recruitment strategies.

The analysis suggests that attaching xPass values to game states can have multifaceted applications. It can facilitate comprehensive performance analysis, enabling teams to understand their playing style variations based on different game situations. Furthermore, it provides a lens for dissecting the strategies of opposing teams, aiding in formulating effective game plans.

A notable implication lies in the realm of player assessment. By delving into individual player’s xPass behaviors across game states, insights can be gained into their adaptability, decision-making, and contribution to team tactics. This, in turn, could be invaluable in recruitment processes, where the observed patterns can guide the selection of players that align with specific team requirements and philosophies.

Perhaps the most intriguing prospect lies in the realm of managerial assessment. The analysis suggests that managers’ influence on a team’s playing style and response to different game situations could be discerned through their team’s xPass behaviors. This opens doors for evaluating a manager’s ability to instill a consistent style and his/her capacity to adapt strategies based on varying game circumstances.

In light of these possibilities, the author expresses enthusiasm for delving deeper into the managerial aspect and intends to expand on this theme in a dedicated future blog post.

In essence, the analysis establishes the initial groundwork for utilizing xPass in conjunction with game states to extract valuable insights about team dynamics, player behavior, and managerial impact. It fosters a call for collaborative exploration to refine and extend the methodology, ultimately providing a tool that enhances the nuanced understanding of soccer strategies, tactics, and decision-making processes.

Be a Team Player — Pass It On!