Research on Amazon's stock price forecasting based on arbitrage pricing model based on big data

Abstract

The generation of big data is based on the network data generated when people use Internet information systems to interact. Big data can reflect the general laws of specific fields and industries, provide more accurate references for decision makers and managers, and provide people with better Data services. Arbitrage pricing model has long been widely quoted by scholars as an alternative theory to capital asset pricing model, which is used to make a regression analysis on Amazon's stock price in this study. In our study, we aim to construct an arbitrage pricing model to make a regression analysis on Amazon's stock price, which is demonstrated to have a higher prediction accuracy and better fitting degree compared with the self-coding network. First, six relevant indicators are selected to conduct establishment of arbitrage pricing model. Then, a self-coding neural network is constructed to estimate the influence coefficients of each factor on Amazon's stock price, which are compared with the results obtained by regression analysis. Finally, the following conclusions are obtained that the arbitrage pricing model has a high prediction accuracy for Amazon's stock price, and the fitting degree of the final model can reach 0.996, which is better than that of the self-coding network.