A
Which soccer metrics best predict winning? A data-driven analysis across Europe’s top five leaguesBishreltBat-Erdene1
AdamPeloguin1
1
A
A
A
Newton North High SchoolMAUSABishrelt Bat-Erdene1 and Adam Peloguin1
1Newton North High School, MA, USA
ABSTRACT
This study tests which team-level metrics best predict success across Europe’s top five soccer leagues. Using data from the 2023–24 and 2024–25 seasons, I analyzed outcomes (goals, goal difference, points), expected metrics (xG, xGA, xGD), style indicators (possession, progression, G + A/90, xG + xAG/90), and a custom finishing efficiency measure. All work was done in Microsoft Excel using correlations, multiple regression, clustering, and residual analysis. Points per match (Pts/MP) was used to standardize success across leagues.
Goal difference and expected goal difference (xGD) had the strongest relationships with Pts/MP. Chance creation metrics (xG + xAG/90 and G + A/90) were also strongly related to results in both seasons. Regression models showed defense mattered as much as attack: lower xGA consistently predicted more points. Finishing efficiency was a useful separator of elite and mid-table teams. Clustering revealed five stable play styles (high-possession progressors, controlled buildup teams, vertical creators, deep-block survivalists, and direct counters), with similar performance gaps in both seasons. At the league level, the Premier League combined higher chance creation with strong results, while Serie A and La Liga achieved similar points with fewer chances; the Bundesliga and Ligue 1 underperformed relative to chance creation.
Overall, success in elite soccer comes from a balance of chance creation, defensive strength, and clinical finishing. Beyond describing team outcomes, the Excel-based workflow also shows how data can reveal consistent tactical identities across leagues and seasons. This makes the approach useful not only for comparing teams, but also for highlighting where strategies succeed or fail in different competitive environments.
A
Introduction
The rise of soccer analytics has transformed how teams, analysts, and fans understand performance. Predicting success in soccer remains complex. Traditional statistics such as goals, shots, and possession provide some insight, yet they often fail to capture the full picture of performance. In recent years, soccer analytics has expanded rapidly, supported by publicly available data and advanced metrics (Anderson & Sally, 2013). Advanced metrics such as expected goals (xG) and progression statistics now allow deeper evaluation of attacking and defensive efficiency (FBref.com, 2025; FBref.com, n.d.; MLSSoccer.com, 2020), yet there remains debate about which indicators most reliably predict success across leagues and seasons (Mackenzie & Cushion, 2013; Sarmento et al., 2014).
Previous studies have often focused on single competitions or short time frames, limiting generalizability. For example, research on the Premier League has emphasized possession and chance creation, while studies in Serie A and La Liga highlight defensive efficiency and tactical compactness (Collet, 2013; Rathke, 2017). Few analyses have compared metrics across multiple leagues using standardized data, leaving open the question of which performance indicators best capture success in different competitive contexts (Liu et al., 2016; Pollard & Reep, 1997; Sarmento et al., 2014).
This study aims to address that gap by analyzing team-level performance across Europe’s top five leagues (Premier League, La Liga, Bundesliga, Serie A, Ligue 1) over two consecutive seasons (2023–24 and 2024–25). We evaluate both traditional and advanced key performance indicators (KPIs) using consistent methods within Microsoft Excel. Our approach combines correlation, regression, clustering, and residual analyses to identify which metrics most strongly predict points per match (Pts/MP), a standardized measure of success. By comparing across leagues and tactical styles, we aim to provide a clear framework for understanding the drivers of team performance in modern soccer.
Methods
Data collection
I downloaded team-level performance stats from the 2023–24 and 2024–25 seasons across Europe’s top five soccer leagues (Premier League, La Liga, Bundesliga, Serie A, Ligue 1) from fbref.com. Raw tables (matches played, goals for/against, expected goals [xG], possession, progressive passes, assists, etc.) were imported into Microsoft Excel and standardized across leagues. In total, the dataset included 98 unique teams across both seasons. All metrics were aggregated at the full-season level (FBref.com, 2025).
Key performance indicators (KPIs)
We analyzed both traditional and advanced KPIs, grouped into four categories:
Outcome metrics: Goals For (GF), Goals Against (GA), Goal Difference (GD), Points, and Points per Match (Pts/MP).
Expected metrics: Expected Goals (xG), Expected Goals Against (xGA), Expected Goal Difference (xGD), and Expected Goals + Expected Assists per 90 minutes (xG + xAG/90).
Style metrics: Possession (%), Progressive Passes (PrgP), Goals + Assists per 90 (G + A/90), xG per Possession, and Progressive Passes per Possession.
Efficiency metric: To measure how well clubs converted expected chances into actual output, we created a custom Finishing Efficiency metric:
A full reference table of all KPIs used in this study is provided below (Table 1).
KPI | Category | Description |
|---|---|---|
GF | Outcome | Goals For (total) |
GA | Outcome | Goals Against (total) |
GD | Outcome | Goal Difference (GF – GA) |
Pts | Outcome | Total Points in the season |
Pts/MP | Outcome | Points per Match |
xG | Expected | Expected Goals |
xGA | Expected | Expected Goals Against |
xGD | Expected | Expected Goal Difference |
xG + xAG/90 | Expected | Expected Goals plus Expected Assists per 90 minutes |
Poss | Style | Possession Percentage |
PrgP (Progression) | Style | Total Progressive Passes |
G + A/90 | Style | Goals plus Assists per 90 minutes |
xG per Poss | Style | Expected Goals per unit of Possession % |
Prog Passes per Poss | Style | Progressive Passes per unit of Possession % |
Finishing Efficiency | Efficiency | Ratio of actual output (G + A per 90) to expected output (xG + xAG per 90) |
Derived metrics (Excel formulas)
Key variables were calculated directly in Excel:
Pts/MP: =Total Points / Matches Played
Finishing Efficiency: (Goals + Assists per 90) / (xG + xAG per 90)
Percentile ranking: =PERCENTRANK.INC(Range, Cell) to rank teams within distributions
Quartile thresholds: =PERCENTILE.INC(Range, 0.25) and = PERCENTILE.INC(Range, 0.75) defined cutoffs for “Low,” “Mid,” and “High” buckets
Efficiency bucket formula (example): =IF($B2 < 0.97,"Low",IF($B2 < = 1.02,"Mid","High"))
Style labels: Possession, progression, and efficiency buckets were concatenated (e.g., =BH2&"-"&BI2) to generate interpretable team profiles such as High-possession progressors or Direct counters.
Statistical analyses
I used MS Excel for the following multiple analyses to assess relationships between KPIs and team success (measured by Pts/MP):
Correlation: =CORREL(Y-range, X-range) tested pairwise KPI associations.
Multiple regression: Conducted using Excel’s Regression tool within the Data Analysis add-in. Outputs included coefficients, standard errors, t-statistics, p-values, and R² values. Residuals were calculated by subtracting predicted Pts/MP from observed values.
Clustering: Teams were grouped into five stylistic clusters (high-possession progressors, controlled buildup teams, vertical creators, deep-block survivalists, and direct counters) using percentile thresholds for possession, progression, and efficiency.
Residual analysis: Compared observed vs. expected Pts/MP to identify over- and under-performers relative to KPI-based models.
Excel analysis tools
Pivot tables: Aggregated averages (e.g., Pts/MP, possession, efficiency) by league and cluster.
Charts: Scatterplots (e.g., xG + xAG/90 vs Pts/MP, with bubble size = possession, color = efficiency) and residual plots. Team names were overlaid using Format Data Labels → Value from Cells.
Conditional formatting: Applied to highlight efficiency levels and cluster differences in visual interpretation.
Results
Elite teams consistently turned dominance into points
Points per match (Pts/MP) was used as the main point of reference across both seasons, since it standardizes team success despite leagues having different total match counts across the top five leagues. Values ranged from 0.32 (Southampton, 2024–25) to 2.65 (Leverkusen, 2023–24). The clubs at the top of each league table were those with the highest Pts/MP, showing a direct link between this metric and league position.
Season | Highest Pts/MP (Elite club) | Pts/MP | Lowest Pts/MP (Relegation club) | Pts/MP |
|---|---|---|---|---|
2023–24 | Leverkusen (Bundesliga) | 2.65 | Granada (La Liga) | 0.55 |
2024–25 | Liverpool (Premier League) | 2.21 | Southampton (Premier League) | 0.32 |
This comparison highlights the gap between elite and struggling teams. While playing styles differed, Pts/MP clearly separated top performers from relegation-threatened clubs.
Chance creation was the strongest predictor of success
When comparing different key performance indicators (KPIs) to points per match (Pts/MP), both seasons showed that creating and converting chances mattered more than possession or efficiency alone.
In 2023–24, the strongest correlations with Pts/MP were goal difference (r = 0.97), expected goal difference (xGD, r = 0.91), and chance creation measured by G + A per 90 (r = 0.85) and xG + xAG per 90 (r = 0.83).
In 2024–25, the pattern held: goal difference (r = 0.97) and xGD (r = 0.93) were the best predictors, with G + A per 90 (r = 0.87) and xG + xAG per 90 (r = 0.84) also strongly matching success.
Possession and progression showed moderate but meaningful correlations (r = 0.73–0.83), while finishing efficiency was relatively weaker (r = 0.50).
These results show that teams consistently generating high-quality chances and maintaining strong expected goal differences were much more likely to succeed, regardless of possession dominance.
KPI | Correlation_with_Pts/MP | |
|---|---|---|
2023-24 season | 2024-25 season | |
GD | 0.97 | 0.97 |
xGD | 0.91 | 0.93 |
G + A/90 | 0.85 | 0.87 |
PrgP (Progression) | 0.83 | 0.81 |
Poss | 0.78 | 0.73 |
xG + xAG/90 | 0.83 | 0.84 |
Finishing_Efficiency | 0.50 | 0.50 |
Efficiency separated the champions from the rest
Regression models were used to test which KPIs predicted success, measured by Points per Match (Pts/MP). Across both seasons, Goal Difference (GD) had the strongest simple correlation with Pts/MP (r = 0.97), but regression models were needed to see which advanced metrics still mattered when considered together.
In 2023–24, the clearest predictor was defensive strength. Lower expected goals against (xGA) was highly significant (p < 0.001), meaning teams that allowed fewer quality chances were consistently near the top. When xGA was removed from the model, ball progression (PrgP) became significant (p = 0.034), and possession showed a borderline effect (p = 0.075). Finishing efficiency was not significant in either model.
In 2024–25, finishing efficiency emerged as an important factor. Teams that converted chances more effectively gained a measurable advantage (p = 0.046 with xGA; p = 0.038 without xGA). Progression was again significant (p = 0.046 with xGA; p = 0.003 without xGA), and xGA remained highly predictive (p < 0.001). Chance creation (xG) trended toward significance in the model without xGA (p = 0.068), but did not fully cross the threshold.
These results suggest that while creating chances is necessary, the clubs at the very top were separated by defensive stability and efficient finishing. Liverpool (2024–25) and Leverkusen (2023–24) both combined those traits, while underperformers such as Manchester United and Sevilla failed to turn solid KPI profiles into consistent points.
Season | Significant predictors (p < 0.05) | Notes |
|---|---|---|
2023–24 | xGA (lower values = stronger defense), PrgP (Progression) | Defensive solidity was the strongest predictor; progression added some value, but finishing efficiency was not significant. |
2024–25 | xGA (lower values = stronger defense), PrgP (Progression), Finishing Efficiency | Defense remained crucial, but finishing efficiency also separated top clubs from others. |
Full regression outputs with coefficients, t-statistics, and p-values are provided in the Excel file.
Finishing efficiency was the great divider
To measure how well clubs converted expected chances into actual output, I created a custom Finishing Efficiency metric:
A value above 1.0 indicates clinical conversion (more goals/assists than expected), while values below 1.0 reflect underperformance. This measure was necessary because traditional xG-based measures capture chance quality but not whether teams consistently finished those chances (Davis & Robberechts, 2024).
Across both seasons, teams with finishing efficiency above 1.0 regularly outperformed expectations. Liverpool, Real Madrid, Arsenal, and Girona turned their chance creation into top point totals. In contrast, Sevilla, Burnley, and Granada underperformed, leaving them in mid-table or relegation battles despite similar xG profiles.
A
Fig. 1
Finishing efficiency and team success across Europe’s top five leagues.
(A) Scatterplots of finishing efficiency vs. points per match (2023–24 and 2024–25). Teams above the 1.0 line were efficient finishers; those below underperformed relative to their chance quality. (B) Top five and bottom five teams by finishing efficiency in each season, highlighting how clinical finishing separated elite performers from struggling clubs.
Five clear playstyle clusters emerged across Europe
Teams across both seasons were grouped into five distinct clusters based on possession, progression, and efficiency:
1.
High-possession progressors (e.g., Man City, Real Madrid, PSG, Bayern) – these clubs consistently paired heavy possession with elite finishing, averaging ~ 2.1 Pts/MP and ranking near the top of their leagues.
2.
Controlled buildup teams (e.g., Napoli, Roma, Lille, Bologna) – patient and balanced possession produced solid but not elite outcomes, averaging ~ 1.7 Pts/MP.
3.
Vertical creators (e.g., Atalanta, Monaco, Atlético) – direct forward play produced mid-table success (~ 1.8 Pts/MP).
4.
Deep-block survivalists (e.g., Dortmund, Newcastle, Lyon, Aston Villa) – defensive setups averaged just ~ 1.2 Pts/MP, with many teams fighting to stay in mid-table.
5.
Direct counters (e.g., Everton, Union Berlin, Cagliari, Ipswich Town) – reactive low-possession play was the least effective, averaging ~ 0.9 Pts/MP and league ranks close to relegation.
Cluster / Style | Avg Pts/MP | Avg League Rank | Notes | ||
|---|---|---|---|---|---|
2023–24 | 2024–25 | 2023–24 | 2024–25 | ||
Controlled buildup teams | 1.7 | ~ 1.7 | 5.3 | ~ 6.0 | Elite: Bologna, Napoli; Mid-table: Roma, Lille, Real Sociedad |
Deep-block survivalists | 1.2 | ~ 1.2 | 11.4 | ~ 11.0 | Mid-table: Dortmund, Villa, Newcastle; Relegation: Lecce, Granada, Monza |
Direct counters | 0.9 | ~ 0.9 | 14.8 | ~ 14.5 | Relegation: Everton, Cagliari, Union Berlin, Ipswich |
High-possession progressors | 2.1 | ~ 2.1 | 3.8 | ~ 3.5 | Elite: Man City, Arsenal, Liverpool, Real Madrid, Barcelona, Bayern, Leverkusen, Inter, Chelsea |
Vertical creators | 1.8 | ~ 1.8 | 5.0 | ~ 5.0 | Mid-table: Atalanta, Monaco, Atlético, Lens |
The same five clusters appeared in both seasons, and their performance gaps were nearly identical. This consistency highlights that the framework captures stable tactical identities across Europe, rather than short-term variation.
A
Fig. 2
Cluster styles and average performance across Europe’s top five leagues (A: 2023–24 and B: 2024–25).
Bar and line charts show average Pts/MP (bars) and average league rank (line) for each of the five clusters. High-possession progressors consistently dominated, while direct counters struggled at the bottom.
Some clubs beat the numbers, others fell short
Residual analysis compared actual performance (Pts/MP) with predicted values from regression models. This highlighted which clubs gained more or fewer points than expected based on their KPI profiles.
In 2023–24, Juventus, Inter, and Atlético Madrid were among the top overperformers, consistently earning more points than predicted. Brest and Nice also stood out in Ligue 1 as teams that converted efficient play into stronger results. On the other hand, Almería and Burnley were clear underperformers, and even Bayern Munich earned fewer points than expected despite strong metrics.
In 2024–25, Napoli recorded the highest positive residual, showing that their results went well beyond what the models predicted. Freiburg and Fiorentina also outperformed expectations, while Roma and Rayo Vallecano added further surprises. By contrast, Tottenham fell far below its predicted values, making it the biggest underperformer. Saint-Étienne and Rennes in Ligue 1, and Holstein Kiel in the Bundesliga, also underachieved relative to their underlying metrics.
This analysis highlights how intangibles like coaching, mentality, and squad balance can push teams above or below their statistical expectations.
Season | Top 5 Overperformers (highest residuals) | Residual | Top 5 Underperformers (lowest residuals) | Residual |
|---|---|---|---|---|
2023–24 | Juventus (Serie A) | + 0.52 | Almería (La Liga) | –0.45 |
Nice (Ligue 1) | + 0.45 | Burnley (Premier League) | –0.40 | |
Inter (Serie A) | + 0.41 | Bayern Munich (Bundesliga) | –0.38 | |
Atlético Madrid (La Liga) | + 0.37 | Salernitana (Serie A) | –0.38 | |
Brest (Ligue 1) | + 0.35 | Darmstadt 98 (Bundesliga) | –0.37 | |
2024–25 | Napoli (Serie A) | + 0.61 | Tottenham (Premier League) | –0.65 |
Freiburg (Bundesliga) | + 0.32 | Holstein Kiel (Bundesliga) | –0.44 | |
Fiorentina (Serie A) | + 0.31 | Saint-Étienne (Ligue 1) | –0.41 | |
Rayo Vallecano (La Liga) | + 0.30 | Rennes (Ligue 1) | –0.39 | |
Roma (Serie A) | + 0.30 | Ipswich Town (Premier League) | –0.36 |
League style profiles
League-wide comparisons showed distinct stylistic differences across Europe’s top five leagues. Table 7 and Fig. 3A-B summarize how finishing efficiency, expected goals plus expected assists (xG + xAG), possession, and results (Pts/MP) aligned.
In 2023–24, the Premier League led in attacking production (2.83 xG + xAG/90) and possession (51.4%), and also achieved the highest points per match (1.52). La Liga and Serie A had lower xG + xAG (2.29 and 2.19) but maintained similar points per match (1.48 and 1.46), suggesting greater efficiency in turning fewer chances into results. Ligue 1 lagged, with both lower finishing efficiency (0.94) and fewer points (1.40). The Bundesliga created many chances (2.66 xG + xAG) but returned a modest 1.42 Pts/MP, indicating underperformance relative to output.
In 2024–25, the Premier League again topped both chance creation (2.65 xG + xAG) and results (1.53 Pts/MP). Serie A and La Liga converged at 1.48 Pts/MP despite Serie A producing fewer chances (2.17 vs. 2.31 xG + xAG), again pointing to greater efficiency. Ligue 1 improved slightly (1.45 Pts/MP) but remained behind, while the Bundesliga stayed the least efficient, averaging just 1.41 Pts/MP despite high xG + xAG.
Taken together, these patterns highlight two consistent themes: (1) the Premier League combined volume (chances and possession) with results, while (2) Serie A and La Liga showed efficiency in converting fewer chances into similar points. By contrast, the Bundesliga and Ligue 1 underperformed relative to their chance creation.
Season | League | Finishing efficiency | xG + xAG/90 | Poss (%) | Pts/MP |
|---|---|---|---|---|---|
2023–24 | Bundesliga | 1.04 | 2.66 | 50.3 | 1.42 |
La Liga | 0.99 | 2.29 | 50.9 | 1.48 | |
Ligue 1 | 0.94 | 2.36 | 50.1 | 1.40 | |
Premier League | 1.01 | 2.83 | 51.4 | 1.52 | |
Serie A | 1.01 | 2.19 | 50.7 | 1.46 | |
2024–25 | Bundesliga | 1.05 | 2.53 | 50.3 | 1.41 |
La Liga | 0.97 | 2.31 | 50.8 | 1.48 | |
Ligue 1 | 0.97 | 2.59 | 50.3 | 1.45 | |
Premier League | 1.01 | 2.65 | 50.9 | 1.53 | |
Serie A | 1.03 | 2.17 | 50.9 | 1.48 |
Fig. 3
League style profiles (A: 2023–24, B: 2024–25).
Bubble charts showing average league styles. x-axis: xG + xAG per 90; y-axis: points per match (Pts/MP). Bubble size = possession; color = finishing efficiency. The Premier League consistently combined the highest chance creation with strong results. Serie A and La Liga achieved similar points with fewer chances, while the Bundesliga generated high xG but underperformed. Ligue 1 trailed in both seasons.
Discussion
This study demonstrates that chance creation, defensive solidity, and finishing efficiency are the primary drivers of success across Europe’s top five leagues. Goal difference unsurprisingly correlated almost perfectly with points per match, but more detailed analyses revealed consistent patterns that extend beyond outcomes alone.
First, both goal difference and expected goal difference (xGD) were the strongest predictors of team success, each showing very high correlations with points per match. In addition, chance creation metrics such as xG + xAG per 90 and G + A per 90 also strongly matched performance, showing that teams that consistently generated high-quality chances were best positioned to succeed. This supports prior work linking passing sequences and shot creation to match outcomes (Hughes & Franks, 2005; Rathke, 2017). This also aligns with prior research emphasizing expected goals as a measure of attacking strength, and our results confirm that combining xG with assists (xAG) provides added predictive value (Sarmento et al., 2014; Mackenzie & Cushion, 2013).
Second, regression analyses showed that defense was equally decisive. Lower expected goals against (xGA) consistently predicted higher points per match, confirming that elite clubs separate themselves not only by scoring but also by limiting the quality of chances conceded. Liverpool (2024–25) and Leverkusen (2023–24) exemplified this balance, while underperforming teams such as Sevilla and Manchester United struggled to convert solid attacking metrics into results due to defensive instability.
Third, finishing efficiency emerged as a key differentiator. While not the strongest single predictor, it consistently explained why certain clubs exceeded or fell short of their expected values. Teams such as Real Madrid, Arsenal, and Girona outperformed models due to clinical finishing, while Burnley and Granada failed to convert chance creation into points. This suggests that efficiency, though variable, remains critical for separating champions from mid-table sides (Davis & Robberechts, 2024).
Tactical clustering reinforced these findings. High-possession progressors dominated across both seasons, consistently averaging over 2 points per match. Direct counters, by contrast, struggled with fewer than 1 point per match on average. The stability of these five stylistic groups across consecutive seasons indicates that the framework captures enduring tactical identities rather than short-term fluctuations (Pollard & Reep, 1997), consistent with earlier work showing that the value of possession varies with situational factors such as scoreline and opposition quality (Collet, 2013; Lago-Peñas & Dellal, 2010).
Finally, league-level comparisons revealed meaningful differences in style. The Premier League combined the highest attacking volume with the strongest results, while Serie A and La Liga demonstrated efficiency by converting fewer chances into similar outcomes. In contrast, the Bundesliga and Ligue 1 generated many chances but underperformed relative to expectations, suggesting structural or tactical inefficiencies. These cross-league differences emphasize that the relationship between metrics and success is shaped by broader competitive environments (FBref.com, 2025; Liu et al., 2016).
Overall, our results suggest that success in elite soccer is best explained by a combination of chance creation, defensive stability, and clinical finishing. These findings provide both a framework for academic analysis and practical insights for coaches, analysts, and recruitment departments.
Conclusion
This analysis of Europe’s top five soccer leagues across two seasons shows that team success depends on more than possession or attacking volume. The strongest predictors of winning were expected goal difference (xGD), chance creation (xG + xAG/90 and G + A/90), defensive solidity (xGA), and finishing efficiency. Together, these metrics explain why elite clubs consistently outperform others and why certain teams exceeded or fell short of statistical expectations.
By applying a simple, reproducible Excel-based framework, we show that tactical styles and league-wide differences can be systematically compared. While high-possession progressors consistently dominated, defensive and efficient teams in Serie A and La Liga also achieved strong outcomes. These results highlight that winning in modern soccer requires not only creating chances but also defending effectively and finishing clinically.
Limitations
This study has several limitations. All data came from fbref.com tables, which may not capture every tactical detail. Analyses were conducted only in Microsoft Excel, without advanced modeling, and the focus was limited to team-level metrics across just two seasons. Player-level variation, coaching, and broader contextual factors such as scheduling and finances were not considered.
A
A
A
Acknowledgements
I would like to thank my advisor and AP Statistics teacher, Mr. Adam Peloguin, for his valuable guidance on this project. I am also grateful to Newton North High School for its academic support and encouragement, and to my family for their constant support.
References
Anderson C, Sally D (2013) The numbers game: Why everything you know about soccer is wrong. Penguin Books
Collet C (2013) The possession game? A comparative analysis of ball retention and team success in European and international football, 2007–2010. J Sports Sci 31(2):123–136. https://doi.org/10.1080/02640414.2012.727455
Davis J, Robberechts P (2024) Biases in expected goals models confound finishing ability. arXiv preprint arXiv:2401.09940. https://arxiv.org/abs/2401.09940
FBref.com (2025) Team statistics for Europe’s top five leagues, 2023–24 and 2024–25 seasons. Retrieved August 2025, from https://fbref.com/en/
FBref.com. (n.d.). Expected goals (xG) explained. Retrieved (2025) August from https://fbref.com/en/expected-goals-model-explained/
Hughes M, Franks I (2005) Analysis of passing sequences, shots and goals in soccer. J Sports Sci 23(5):509–514. https://doi.org/10.1080/02640410410001716779
Lago-Peñas C, Dellal A (2010) Ball possession strategies in elite soccer according to the evolution of the match-score: The influence of situational variables. J Hum Kinetics 25(1):93–100. https://doi.org/10.2478/v10078-010-0036-z
Liu H, Hopkins W, Gómez MÁ (2016) Modelling relationships between match events and match outcome in elite football. Eur J Sport Sci 16(5):516–525. https://doi.org/10.1080/17461391.2015.1042527
Mackenzie R, Cushion C (2013) Performance analysis in football: A critical review and implications for future research. J Sports Sci 31(6):639–676. https://doi.org/10.1080/02640414.2012.746720
MLSSoccer.com. (2020), April 16 Soccer analytics 101: Expected goals. https://www.mlssoccer.com/news/soccer-analytics-101
Pollard R, Reep C (1997) Measuring the effectiveness of playing strategies at soccer. Stat 46(4):541–550. https://www.jstor.org/stable/2988603
Rathke A (2017) An examination of expected goals and shot efficiency in soccer. J Hum Sport Exerc 12(2):514–529. https://doi.org/10.14198/jhse.2017.122.15
Sarmento H, Marcelino R, Anguera MT, Campaniço J, Matos N, Leitão JC (2014) Match analysis in football: A systematic review. J Sports Sci 32(20):1831–1843. https://doi.org/10.1080/02640414.2014.898852