— In Content Delivery Networks (CDNs), the knowledge of the popularity of video contents helps the manager to take efficient decisions about which video contents should be cached near the end users and also about the duplication degree of each video to satisfy the end user Quality of Experience. This paper focuses on predicting the popularity of video contents, in terms of number of solicitations. For that purpose, different software entities, called experts compute popularity values of video contents. Each expert uses its own prediction method. The expert prediction accuracy is evaluated by a loss function as the discrepancy between the prediction value and the real number of solicitations. The simulations based on real YouTube traces show that the accuracy of prediction is improved by splitting the video content profile in contiguous phases. Different prediction methods are compared and also different phase change-points detection methods are evaluated in order to identify the method (or method parameters) minimizing the cumulated discrepancy compared to real solicitations of video contents.