Alphanetv4: Alpha Mining Model
Downloads
As AI and deep learning have become hot spots in the 21st century , they are widely used in the current quant market. In 2020, Huatai Securities constructed deep-learning-based AlphaNet for stock feature extraction and price prediction. At present, it has developed to the 3rd version and has formed a great influence in the market.
However, the AlphaNet has some problems, such as underfitting caused by short sequence length of feature extraction, insufficient diversity of feature extraction, high complexity, instability of random sampling, which lead to the poor performance. So this paper proposes AlphaNetV4 to solve them. The main contributions of this paper are: 1) Increased the length of the sequence and reduced the step size of the extraction layer to improve the fitting effect; 2) Reduced the relevance of original input; 3) Used Spearman correlation coefficient to design dropout layer instead of random sampling to enhance the stability of feature extraction; 4) Applied Bi-LSTM to enrich the extraction layer, and Transformer to enhance the learning ability of the model. In addition, this paper also uses CNE5 Barra to redesign the fitting target, and optimizes the training process by modifying the training weight and using sharp EarlyStopping. This paper compares the performance between AlphaNetV4 and the previous AlphaNets. It verifies that increasing the sequence length can reduce the loss from 0.5 to 0.3, reducing the correlation of input can reduce the loss to 0.25, using Spearman Dropout can cut the computational complexity without damaging the accuracy, and that Transformer can reduce the loss to less than 0.1. Further, this paper conducts the back test to show that AlphaNetV4 has increased the annual excess return by about 7% - 10%. Finally, this paper provides suggestions on the future development of quant trading.
Downloads
1. White Economic prediction using neural networks: The case of IBM daily stock returns[C]. IEEE 1988 International Conference on Neural Networks, 1988.
2. Diler AI Predicting direction of ISE national-100index with back propagation trained neural network[J].Istanbul Stock Exchange, 2003, 7(25-26): 65-82.
3. Dong Jianfang. The current status and development trend of deep learning in quantitative investment at home and abroad[J]. Modern Business, 2022(01):120-122.DOI:10.14097/j.cnki.5392/2022.01.038.
4. Cui Jianfu, Li Xingxu. Stock price prediction: comparison between GARCH model and BP neural network model[J]. Statistics and Decision, 2004(06):21-22.
5. Xin Zhiyun, Gu Ming. Complex financial time series forecasting based on least squares support vector machine[J]. Journal of Tsinghua University (Science and Technology). Preview, 2008(07):1147-1149.
6. Yao Xiao, Yu Le'an. Fuzzy approximate support vector machine model and its application in credit risk assessment[J]. Systems Engineering - Theory & Practice, 2012, 32(03): 549-554.
7. Su Bing. Quantitative timing research based on PCA-SVM model[D]. Tianjin University of Finance and Economics, 2015.
8. Alex,Ilya Sutskever,Geoffrey. Hinton. ImageNet classification with deep convolutional neural networks[J]. Communications of the ACM,2017,60(6):
9. Hochreiter,Schmidhuber. Long short-term memory.[J]. Neural computation,1997,9(8) :
10. Dzmitry Bahdanau,Kyunghyun Cho,Yoshua Bengio. Neural Machine Translation by Jointly Learning to Align and Translate.[J]. CoRR,2014,abs/1409.0473:
11. Mike Schuster,Kuldip Paliwal. Bidirectional Recurrent Neural Networks[J]. IEEE Transactions on Signal Processing: A publication of the IEEE Signal Processing Society,1997,45(11):
12. Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun 0001. Deep Residual Learning for Image Recognition.[J]. CoRR, 2015, abs/1512.03385:
13. Liang Jiaying. Research on stock selection strategy based on MSCI Barra CNE5 model style factors[D]. South China University of Technology, 2020.DOI:10.27151/d.cnki.ghnlu.2020.004068.
14. Tang Cheng. Quantitative investment research based on Attention-LSTM deep learning method[J]. Business News, 2021(36):155-157.
15. Lin Xiaoming. AlphaNet: Factor Mining Neural Network. Huatai Securities 2 020 .
16. Lin Xiaoming. Revisiting AlphaNet: Structure and Feature Optimization. Huatai Securities 2020.
17. Lin Xiaoming. AlphaNet launches INSIGHT platform. Huatai Securities 2 021 .
18. Lin Xiaoming. AlphaNet Improvement : Structure and Loss Function. Huatai Securities 2 021 .
19. Li Junqing. Comprehensive safety analysis system based on factory power controller data[D]. Shanghai University of Technology, 2019.DOI:10.27801/d.cnki.gshyy.2019.000308.
20. Vaswani, Shazeer, Parmar, et al. Attention is all you need[J]. Advances in neural information processing systems, 2017, 30.
21. Wang Yan. Research on Arbitrage in CSI 300 Index Futures Market[D]. East China University of Political Science and Law, 2013.
Copyright (c) 2022 Wenjun Wu
This work is licensed under a Creative Commons Attribution 4.0 International License.