2020
MCM/ICM
Summary Sheet
Team Control Number
2003723
Teaming Strategies in Football: Patterns and Effects
Objective analysis of the football team's performance is the main way to set training targets, and
improve the team's level. Existing analysis of team performance is mostly based on the
traditional technical statistics of individual athletes, but seldom discusses the important roles of
the interrelationships and synergies between athletes. In view of this, this paper adopts the
perspective of players' cooperation network, constructs the analysis framework of "identifying
cooperation network–expanding cooperation model-exploring influence effect–putting forward
suggestions", quantitatively analyzes the cooperation relationship of Huskies football team and
its influence law on competition performance. We have mainly solved the following four
problems:
Firstly, in order to identify the cooperation network, this paper constructs a complex network
model based on directional weighted graph based on the characteristics and data of passing the
ball. This paper constructs cooperation network indicators and other structural indicators from
multiple scales. We refer to gene coding types to distinguish binary and ternary structures, use
the simulation chart of competition field to mine the team formation rules, and adopt
heterogeneity analysis to explore the relationship between individual players and the team, as
well as the change rules of different network indicators with time.
Secondly, in order to measure team cooperation performance more comprehensively, this paper
constructs a comprehensive evaluation index system by integrating Individual contributions to
the team, Team structure, Strategic layout, Environment variables and 13 sub-indicators. Among
them, in order to make clear the Individual contribution to the team, this paper combines
subjective and objective weights, uses AHP and entropy weight to construct a sub-model of
player scoring.
Thirdly, we research the mechanism where team strategy and other factors affect victory. For
heterogeneity analysis, we build multi-classification undirected logistic regression models
from two levels, winning and losing situations and score results to distinguish team formations.
We also deal with multiple collinearity and other issues. The results show that Huskies is suitable
to adopt an attack-defense balanced formation and a right-wing attack style, and it is more
effective to adopt counter-strategies appropriately according to the opponent's situation.
Fourthly, according to Huskies' network structure and relevant research conclusions, it is
believed that the overall framework and contingency mechanisms, individual and collective
factors, subjective and objective factors, long-term and short-term situations should be
considered in order to form an excellent football team. Extended to other types of teams, other
factors such as leadership, gender composition and team culture should be considered as well.
To sum up, the advantages of this paper are as follows: (1)Pertinence: choose statistical
indicators and models that can better reflect the characteristics of football matches and player
networks; (2)Clear hierarchy:make different analysis and comparison from the angles of
attack style, team formation, sample range and consider the outcome of the match; (3)Reliability:
use robust analysis method and multiple tests; (4)Extensibility: put forward practical suggestions
to the football team and even the broader competition team.
Key words: Team Strategy, Football, Network Science, Logistic Regression