Universal change point testing for dependent data

One of the main interests in time series analysis is the detection of the so called change-points, defined as timestamps where the model parameters expe- rience a substantial shift in value. Once a candidate change-point is identified, we may want to test whether there is a significant difference in distribution before and after the structural break. In this work we approach the problem from a split-sample perspective and we implement and test on both simulated and real data a two-sample test for time dependent streams that we call universal change-point testing.