The Multiplayer online game industry is growing tremendously. The matchmaking procedure is a very complex task which determines the success and satisfaction of the players. The player's potential is determined by his/her win count and game rating. A team with equivalent potential players makes the game more interesting. Equal skill score competitors have a probability of a 50% success rate.Principal cloud service provider enables online games to gain efficacy and popularity with serverless technology. The expectationMaximization Clustering approach categorizes the players into different levels depending on their ability. Matchmaking among players in the same cluster enhances the quality of multiplayer games. The 0.5 million sample dataset from Kaggle shows only a 2% difference between players who dropped and won the game. In this paper, a matchmaking method is proposed for multiplayer online 5v5 games with minimum drop-risk using ExpectationMaximization clustering on the player pool.