Linear Programming as a Tool for Managing the Training Process of Esports Teams

Abstract

Background. Linear programming, with its ability to account for multiple constraints and optimize linear objective functions, is a promising tool for solving training planning tasks. This method enables the development of individualized plans tailored to the specific goals of each player and the team as a whole. Objectives. This study aimed to develop linear programming variants for automating the planning process of esports teams’ training schedules, enabling the allocation of workloads and determining the optimal distribution of time across various types of training while considering the individual characteristics of esports athletes, constraints, and diverse strategic goals. Materials and methods. A comprehensive analysis of scientific, methodological, and specialized literature was conducted to ascertain the optimal use of resources, expert evaluation methods, linear programming, as well as statistical methods. Reliable statistical methods were employed: the dichotomous scale (results were processed using Cochran’s Q concordance coefficient, which determined the consistency of expert opinions regarding each type of training); and the ratio scale (ranking) – the consistency of opinions was analyzed using Kendall’s W concordance coefficient. Results. An algorithm for determining the distribution of training workloads was proposed, which takes into account expert-defined ratios and constraints. To optimize the planning of the training process for esports teams, linear programming variants were developed, describing the distribution of time between different types of training as a linear programming task. Variant 1 serves for static time allocation between various training types without optimizing the distribution for specific goals. Variant 2 optimizes the time allocation, considering the individual characteristics of athletes and the strategic goals of the team. It incorporates constraints such as the total weekly training hours, the minimum required time for each type of training, and other limitations. The MS Excel Solver linear optimizer was used to find the optimal time distribution. Variant 2 facilitates the investigation of different scenarios for planning the training process of esports teams, demonstrating how the distribution of time across training types changes depending on set goals and preparation phases. Conclusions. Based on the proposed algorithm, linear programming variants were developed, successfully addressing the task of automating the planning of esports team training. In contrast to Variant 1, Variant 2 offers an optimal distribution of time among different types of training (team training, individual training, physical activity sessions, etc.), considering the individual characteristics of players and the strategic goals of the team. It demonstrates high flexibility and adaptability to various esports disciplines, thereby allowing the investigation of different scenarios. The proposed approach can serve as a foundation for creating more advanced systems for managing the training process. Future research prospects include expanding the functionality of linear programming by incorporating additional factors such as psychological aspects, social dynamics within the team, and the physiological indicators of athletes.