SCADA control systems are the keystone for reliable performance optimization of wind farms. Processing into knowledge the amount of information they spread is a challenging task, involving engineering, physics, statistics and computer science skills. The present work deals with the effects on the efficiency of turbine inability of optimal aligning to the wind direction, due to meandering wind caused by wakes. The approach is tested on a judiciously chosen cluster of turbines of a wind farm sited in southern Italy. By a post-processing method based on discretization of nacelle position measurements, a set of dominant patterns of the cluster is identified. The patterns associated to best performances are individuated and it is shown that they correspond to non-trivial alignment to wind direction.