The accurate prediction of wind power is important to guarantee the security and stability of any power system containing a large contribution from wind energy. To improve the accuracy of predictions of wind power generation, many new prediction methods have emerged. However, the accuracy of wind power prediction is not only related to the prediction methods, but also associated with the volatility of wind power. Regardless of the method adopted, it is not possible to guarantee accurate and error-free prediction. The necessity of the volatility in the wind power sequence was explained. To depict the probability of the occurrence of a new volatility mode, based on phase space reconstruction, a recurrence plot, and the recurrence rate was proposed to qualitatively and quantitatively depict the volatility. The changing rules of the recurrence plot and recurrence rate at different spatial scales were discussed. Based on this, a method to analyze the relationship between the volatility of a wind power sequence and prediction error was established. Finally a fair method for evaluating predictions accuracy was provided based on the recurrence rate. The study further illustrates the effectiveness of the method.