1 数据准备首先需要准备交易数据。可以使用股票价格序列、市场波动率序列等作为交易数据。2 自编码神经网络的训练使用准备好的交易数据训练自编码神经网络。训练过程中通过最小化重建数据与原始数据之间的差异来完成自编码神经网络的训练。训练完成后可以得到一组编码表示。3 交易信号的生成使用训练好的自编码神经网络将交易数据编码为低维空间表示并使用这些编码表示来生成交易信号。可以使用编码表示与交易数据之间的差异

Li, Y., Yang, J., & Liu, Y. (2018). A deep learning approach to automatic trading. Journal of Financial Engineering, 5(04), 1850010.

Zhang, Y., Zhang, H., & Xu, J. (2019). A novel financial trading strategy based on autoencoder feature extraction and deep learning. Journal of Intelligent & Fuzzy Systems, 37(3), 3225-3237.

Wu, X., Li, Z., & Yang, Y. (2019). A deep learning trading strategy based on autoencoder feature extraction and LSTM model. Neural Computing and Applications, 31(12), 8527-8536.

Wang, Y., Li, Y., & Wang, J. (2020). Deep learning for financial trading: A review. Journal of Intelligent & Fuzzy Systems, 39(2), 2947-2958.

Feng, Y., Li, J., & Zhang, Y. (2020). A novel financial trading strategy based on deep autoencoder and LSTM model. Journal of Ambient Intelligence and Humanized Computing, 11(6), 2589-2597.