With the development of mobile and GPS technologies, gamers in spatial games based on a real map or a virtual game map play against each other to determine winner and loser of the game. To improve their winning rate, gamers want to play against weak opponents, but it is limited to a certain degree for gamers to choose weak opponents. In general, the game server shows the attributes of other gamers around the current gamer, such as their game level, power, and difficulty. This paper proposes an optimal opponent search technique that allows game servers to search and show possible game opponents around a gamer by comparing their attributes. For comparison of possible game opponents, gamers and their game opponents are evaluated by Euclidean distance. Experiments were performed focusing on searching for game opponents with a danger level equal to or lower than that of the gamer. We measured the similarity between gamers and neighboring opponents based on virtual data to provide an optimal game opponent evaluation that could increase the winning rate of gamers.