The Anatomy of the Global Football Player Transfer Network And how topological properties of the football player transfer network reflect the successfulness of professional football clubs Xiao Fan LIU（刘肖凡） PhD, Associate Professor School of Computer Science and Engineering, Southeast University, Nanjing, China 11/6/2018 Quantifying Success, Seoul, Korea

Quantifying Success, Seoul, Korea 11/6/2018 Quantifying Success, Seoul, Korea

Quantifying Success, Seoul, Korea It always kept me wondering… How is a club’s successfulness affected by its operations in the transfer market? 11/6/2018 Quantifying Success, Seoul, Korea

Successfulness of a professional football club Ultimately measured by its profitability Profit comes from two main sources Good competition results Domestic leagues International competitions Larger commercial revenue Profit from player transfers Direct income Transfer balance Profit from reselling The two measures are not correlated Avg. league pts. 1.7 2.4 IFFHS CWR pts. 753 1519 Transfer balance -14M -25M Reselling profit -16M 3M 11/6/2018 Quantifying Success, Seoul, Korea

Quantifying Success, Seoul, Korea Most successful clubs (only by our standards) 11/6/2018 Quantifying Success, Seoul, Korea

Overview of the transfer market Data collected: 2011~2015 24 major leagues 410 elite clubs More than 10,000 movements Transfer and loan Observation: Increasing transfer expenditure Increasing financial imbalance Money = match win 11/6/2018 Quantifying Success, Seoul, Korea

Quantifying Success, Seoul, Korea Generally, money = win. But that’s not all. How is the clubs’ position in the transfer network affecting its successfulness? 11/6/2018 Quantifying Success, Seoul, Korea

Construction of the player transfer network Nodes: clubs Edges: directed, eij : i→j denotes player transfers from club i to club j General appearance: Small world network, subjected to spatial constrains Almost symmetrical degree distribution No “rich-club” phenomenon 11/6/2018 Quantifying Success, Seoul, Korea

Quantifying Success, Seoul, Korea 11/6/2018 Quantifying Success, Seoul, Korea

Quantifying Success, Seoul, Korea Coreness properties Methods: Eigenvector centrality: PageRank centrality: Aim: To measure the attractiveness of a club to players Remarks: The two centrality measures are all based on number of incoming edges, but have subtle differences. 11/6/2018 Quantifying Success, Seoul, Korea

Quantifying Success, Seoul, Korea Brokerage properties Methods: Effective size： Closeness centrality Betweenness centrality: Aim: To measure the brokerage power of clubs in transfer market. Remarks: Brokerage power indicates the extend of exclusive control of player resources in the transfer market. 11/6/2018 Quantifying Success, Seoul, Korea

Measurements have subtle differences 11/6/2018 Quantifying Success, Seoul, Korea

Comparison of network properties and successfulness Eigenvector PageRank Eff. size Betweenness Closeness League pts. 0.18 0.12 0.34 0.33 CWR pts. 0.36 0.14 0.48 0.37 0.49 Transfer balance 0.04 -0.02 0.19 0.16 0.26 Reselling profit 0.06 0.07 Calculate the concordance of different measures Kendall’s Tau on club rankings Brokerage power plays the most significant role in determining its match outcome; Clubs with higher coreness might win more matches or make more profit from player transfer A good starting point… 11/6/2018 Quantifying Success, Seoul, Korea

Quantifying Success, Seoul, Korea Conclusion The successfulness of professional football clubs can be easily compared with its topological properties in player transfer network. Clubs in different leagues have essentially different recipe of making profit. Further interpretation can shed light on why clubs are successful in actual football world. Read more: “The Anatomy of the Global Football Player Transfer Network: Club Functionalities versus Network Properties”. PLoS One, to appear tomorrow. 11/6/2018 Quantifying Success, Seoul, Korea

Time to shoot your questions Contact me: xfliu@seu.edu.cn 11/6/2018 Quantifying Success, Seoul, Korea

Quantifying Success, Seoul, Korea 国际转会 vs. 国内转会 特征向量 PageRank Eff. size 介数 接近度 联赛平均分 国际转会 0.29 0.19 0.03 0.28 0.36 国内转会 -0.04 -0.06 0.34 0.04 0.15 IFFHS得分 0.40 0.24 0.12 0.48 0.13 -0.05 0.45 0.1 0.23 转会收益 0.05 0.11 0.07 0.16 0.08 转会溢价 0.01 0.02 0.06 -0.01 国际转会子网核心度和比赛成绩具有较强的相关性 国际转会子网全局中介性和比赛成绩具有较强的相关性 国内转会子网本地中介性和比赛成绩具有较强的相关性 转会收益和节点网络特征没有明显关系 11/6/2018 Quantifying Success, Seoul, Korea

Quantifying Success, Seoul, Korea 联赛的财政特征 不同联赛中的俱乐部的转会收益和比赛成绩相关性不同 24个联赛根据这个相关性分类 Money Leagues：转会负收益越大，比赛成绩越好 Farm Leagues：转会正收益越大，比赛成绩越好 Outlier Leagues：转会收益和比赛成绩相关性较弱 分类 包含的联赛 Money leagues England, France, Germany, Italy, Russia, Spain, Turkey, Ukraine Farm leagues Argentina, Belgium, Brazil, Chile, Finland, Greece, Netherland, Norway, Portugal, Romania, Scotland, Sweden Outlier leagues Australia, Ireland, Mexico, USA 11/6/2018 Quantifying Success, Seoul, Korea

Quantifying Success, Seoul, Korea Money vs. Farm Leagues   特征向量 PageRank Eff. size 介数 接近度 联赛平均分 Money 0.25 0.15 0.33 0.29 0.34 Farm -0.01 0.03 0.06 0.04 IFFHS得分 0.30 0.08 0.44 0.39 0.51 0.43 0.53 0.41 转会收益 -0.13 -0.14 -0.09 -0.03 0.35 0.20 0.56 0.46 0.59 转会溢价 -0.04 -0.05 0.19 0.14 0.23 0.17 Money Leagues中俱乐部的比赛成绩与核心度、中介性 都存在相关性 Farm Leagues中俱乐部的国际比赛成绩与和核心度、中 介性都存在相关性 Farm Leagues中俱乐部的转会收益与核心度、中介性都 存在相关性 11/6/2018 Quantifying Success, Seoul, Korea

Quantifying Success, Seoul, Korea 转会网络和租借网络 特征向量 PageRank Eff. size 介数 接近度 联赛平均分 转会 0.18 0.12 0.34 0.33 租借 -0.07 -0.20 0.22 0.28 0.23 IFFHS得分 0.36 0.14 0.48 0.37 0.49 0.11 -0.12 0.43 0.39 转会收益 0.04 -0.02 0.19 0.16 0.26 0.10 0.01 0.13 0.06 转会溢价 0.07 0.05 租借网络和转会网络的构建类似 两个网络中俱乐部核心度与比赛成绩的相关性相差较大 原因可能是：球员的大量转入代表俱乐部有实力，但是 球员的大量租入反而代表俱乐部实例较弱。 11/6/2018 Quantifying Success, Seoul, Korea