If the pattern "is there", the information about it is present in your data set. In my point of view, an approach would be: (i) verify if your data set has enough information (appropriate observability) about the underlying process, (ii) find the appropriate analytical technique (time series analysis, complex system analysis ) , if any, that could capture that information and (ii) look for an appropriate representation to express the patterns found (recurrence plots?). How does your data set looks like?

Hi Leonardo, thanks for the answer. I am gonna answer you in english to make us understandable for more people.

Some researchers has been used the Lag Sequential Analysis, into the Observational Methodology, to acces hidden patterns of soccer games (in other situations they've been used the T-pattern analysis). These patters are presented using the lag method (commonly with 5 prospective or retrospective legs) . For this purpose, it is need to adopt a target and a given criteria. The target criteria is usually the ending of the offensive process (i.e. score a goal, shooting on goal).

However, a soccer game usually has offensive sequences with to many lags, what give us more conditions to find, through the above mentioned tools, these patterns, and a soccer small-sided game allow the players to acces the goal easier and faster, given the small distance between the deffensive field and the opponenent goal. For this reason, usually there is less lags in a offensive sequence in small-sided games than in a 11x11 game.

Conceptually, these patterns exist. Although the game is complex, random and unpredictable, the stablished relationships between players and teams is oriented by the interaction between player-task-enviroment, what makes this context a little predictable. 

In this context, do you know a better tool to find these patterns?