Research On ELO And Matching Algorithms In Competitive Games

With the rapid development of network communication technology and the improvement of people's living standards,as well as the rapid popularization of mobile devices,more and more people begin to play games and games on the network or mobile devices.Nowadays,in many game platforms,such as chess,billiards,landlords and other games,people need an effective ranking system to judge the strength of players or participants in the game or competition.Now the ELO algorithm in chess is the most commonly used ranking algorithm in competitive games.After having a more accurate strength score,the matching system can match according to the player's strength score,so as to help players find the right teammates and opponents.The development of online games also promotes the development of matching system.At present,the matching algorithms commonly used in games are grouping matching algorithm based on strength and grouping matching algorithm based on players'roles.In order to overcome the shortcoming of slow convergence speed of ELO algorithm in competitive games,this paper uses simulation experiment to test the performance of ELO algorithm.Three basic indexes are selected:convergence speed,convergence radius and convergence density.On this basis,the performance of ELO algorithm is tested.The improved ELO algorithm is proposed according to the actual needs.The algorithm introduces the winning threshold and winning reward.The simulation test results show that the convergence speed has been significantly improved compared with the improved ELO algorithm.At the same time,the improved ELO algorithm is applied to the pattern extraction of Go board,and the pattern extraction rules are designed.Two ELO algorithms are used to extract the LCP pattern from the chessboard data respectively,and two sets of ELO scores of fallen pieces under different local chessboard conditions are obtained.The simulation results show that the improved ELO algorithm can extract better LCP pattern.At the same time;in order to optimize the effect of traditional online matching algorithm,the traditional matching algorithm is studied,and greedy matching algorithm and neighborhood planning matching algorithm are proposed.Greedy matching algorithm has three main mechanisms:matching capacity check,replacement mechanism and half-life mechanism.Experiments are expected to prove the advantages of this algorithm compared with traditional grouping matching algorithm.Neighborhood programming algorithm designs the algorithm model according to the matching system design,combines different matching targets into objective functions through different coefficients,and carries on the iterative calculation according to the neighborhood search algorithm,which can effectively improve the matching optimization efficiency,and uses the half-life mechanism of greedy algorithm to constrain the time.In this paper,a domestic MOB A game servers on March 11,2018 stored in more than 2 hours about 2,000 matching data of 5 V5 groups as experimental data,which is used in greedy matching algorithm and neighborhood planning matching algorithm,and compared with the actual online grouping matching algorithm,both of them can obtain better matching results compared with the algorithm used at that time.