Modern coaching of professional kayak paddlers is supported by on-water measurement devices and scientific analysis techniques. However, it is very difficult to determine perfect paddling technique and the interpretation of the measured signals is rather complex. Therefore it is quite challenging to provide easily usable but still accurate indicators of paddlers' performance. In this paper, we show that fluctuation analysis of movement signals provides that can be used to classify the performance of athletes. The standard deviation and Poincaré plots of classical parameters turned out to be useful new indicators of paddling skills, furthermore the signal to noise ratio calculated from power density spectrum of raw movement signals shows relationship with the quality of paddling technique, too. The detailed analysis have been made for paddlings of 26 athletes with different ages and technical skills, which measured by a universal measurement system developed for this task.