a new strategy for stock pricing based on machine learning algorithms

Stock price prediction is a complicated and interesting problem. noisy trends make stock pricing sensitive and complicated while the economical motivation behind, keeps it interesting for researchers and investors. in this paper we are to outline two novel ideas for combining well-known machine learning based methods to provide an improved prediction. the first idea is based on using the best algorithm between the selected choices until its performance decrease during short periods. the second idea is to assign an index to the best method in each day and interpolate the index of the method as a polynomial function of the day index. the interpolation might be used to estimate the index of the best method for each test date. we also test each suggested algorithm for predicting the price of 2 stocks from national market. to show the efficiency of our proposed algorithm, we compare the predicted prices with real values over a test set. moreover, a backtest analysis is performed to verify the closeness of annual returns based on real and predicted prices.