In the present research, we develop a fusion model that is based on the positive combination of LSTM and Transformer models and is aimed at creating a deep learning-based statistical arbitrage strategy. The model was subjected to comparative analysis with both the LSTM model and the statistical cointegration approach, testing it through the Science and Technology Board (STB) dataset during both bull and bear market conditions. A complete analysis of the results shows that the new model with fusion trans_LSTM demonstrates better efficiency, enabling higher returns, and confirms that the overall stock market trends were confirmed during the backtesting of bull and bear market trainings. This quality demonstrates that the models' performance can be judged by the fact that the trans_LSTM fusion model shows effectiveness and reliability in stock market arbitrage. Thereby, it can be seen that this investigation contributes to solutions for further improvement of the role of deep learning in the financial market sphere.

Na Hu, Mingming Qu. A Study of Trans-LSTM Based Arbitrage Strategy for Science and Technology Boards. Academic Journal of Computing & Information Science (2024), Vol. 7, Issue 4: 62-71. https://doi.org/10.25236/AJCIS.2024.070409.

[1] Elliott R J, Hoekd V D, Malcolm W P. Pairs Trading [J]. Quantitative Finance, 2005, 3(5):271-276.

[2] Fischer T, Krauss C. Deep learning with long short-term memory networks for financial market predictions [J]. European Journal of Operational Research, 2017, 270(2).

[3] Zhijing He, Sudan Jin. Realization of pair trading strategy in steel industry based on cointegration theory [J]. Investment and Entrepreneurship, 2020, 31 (23): 14-16.