Stock market forecast is always a hot and difficult point in financial market analysis.Some traditional forecast models are difficult to predict the stock market effectively.In order to solve this problem, two hybrid models of fractal interpolation and SVM, as well as fractal interpolation and BP neural network are proposed by combining fractal interpolation method with machine learning algorithm.The proposed hybrid models firstly use machine learning algorithm to calculate the interpolation points needed for fractal interpolation, and then build a fractal interpolation extrapolation model to predict other values.The empirical results show that the prediction effect of the two hybrid models is better than that of the fractal interpolation model alone, and the prediction accuracy is higher.Therefore, the hybrid model obtained by the combination of fractal interpolation method and machine learning algorithm can better predict non-stationary financial time series such as stock market index.