Author:
Shi Guangde,Gao Jingkai,Li Ruibin,Shi Jun
Abstract
Quantitative trading decision models have a key influence on financial investment. Firstly, this study established an LSTM model by using long-term and short-term memory networks and predicted the future prices of gold and bitcoin investment products. Then, according to the time range of gold and bitcoin assets, three types of transactions were determined: cross, non-cross, and inclusion relationship, and the daily trading strategies were determined by the greedy model established by a greedy algorithm. Then, the Sharpe Ratio of the nonparametric method was used to measure the risk of the developed decision model and evaluate the accuracy of the model. Finally, starting from the stock market fluctuation and macro-mobilization, the sensitivity of the decision model under different transaction costs was tested by increasing or decreasing the percentage of transaction costs (0.5%, 1%, 1.5%, and 2%, respectively). Research informs investors on how to invest for the best returns.
Reference10 articles.
1. X.X. Chen, Research on asset allocation of Chinese investors in global market, Tongji University, Shanghai, 2008.
2. Y. Li, Research on the application of quantitative trading strategy model, Yunnan University of Finance and Economics, Kunming, 2014.
3. J.J. Wang, Quantitative trading applications in China's stock market, Nanjing University, Nanjing, 2013.
4. C. Ravi, Fuzzy Crow Search Algorithm-Based Deep LSTM for Bitcoin Prediction, in: International Journal of Distributed Systems and Technologies (IJDST), IGI Global, Pennsylvania, Hershey, 2020, pp. 53-71.
5. B.B. Sahoo, R. Jha, A. Singh, D. Kumar, Long short-term memory (LSTM) recurrent neural network for low-flow hydrological time series forecasting, in: Acta Geophysica, Springer, Berlin, Heidelberg, 2019, pp. 1471–1481.