Research on stock trading strategy based on deep neural network

Deep neural network is widely concerned with the concept of deep learning. However, there are few researches focus on the application of deep neural network for stock trading strategy. Excellent trading strategies can not only help investors get high profit, but also effectively reduce the income risk. This paper studies 7 trading strategies based on a deep neural network, and uses the 2009-2015 years historical data of Shanghai Composite Index for experiments through sliding window approach, and adopts the accuracy, rate of excess return, volatility of yield and information ratio to measure the advantages and disadvantages of different trading strategies. According to the experimental results, a trading strategy suitable for the deep neural network is found. This trading strategy can not only achieve a high predictive accuracy but also have a low volatility, which can help investors reduce the risk of loss effectively while obtaining satisfactory returns.