While describing a  game, details of game moves in a commentary are not considered, which loses the data. Due to lack of expertise in writing or narrating the game, surprises cannot be revealed completely which loses the interest of game abstract description. The actual spectator has more information about surprises, gameplay, and game conditions in his/her observation of the game.  The surprise-based abstraction of the game represents more information than other descriptions. The writer's perspective, reader’s perspective, and viewer’s perspectives are different which has more information; each one having its perspective for the game surprises. The unrevealed surprises can be mined from each perspective which might create interest in the description. Existing statistical models do not cover these perspectives and surprises. An abstraction of these multi-perspective surprises is proposed in this paper. The surprises are calculated with the help of event uncertainty in the game. The surprising similarity can be measured for various games which can help to decide the game plan, strategy, player style, team psychology. The surprising similarity can be measured with the help of uncertain events in times series of surprises.  The surprise-based similarity is compared with the time series similarity measures.  The surprise-based similarity game comparison has improved. The surprises can also be revealed in other time-series data like stock market data, network traffic data, etc.