Design and Empirical Analysis of Financial Quantitative Trading Model based on VMD-DCNN-SGRU Architecture and Integrated System

In the context of increasingly complex and volatile financial markets, traditional investment decisions cannot effectively deal with massive data and market uncertainties, so it is urgent to optimize investment returns and reduce risks with precise technical means. With the rapid development of financial technology, the innovation of quantitative trading technology has promoted the application of intelligent investment strategies. This paper proposes a quantitative trading model (VMD-DCNN-SGRU) that integrates variational mode decomposition (VMD), deep convolutional neural network (DCNN) and Stacked gated cycle Unit (SGRU), and combines data envelopment analysis (DEA) method to construct investment target screening and decision-making framework. VMD is used for multi-scale decomposition to extract effective features of transaction data, data reconstruction technology enhances structured expression, DCNN strengthens depth features, SGRU realizes timing prediction and outputs investment signals, and DEA method optimizes investment portfolio by combining risk and return. The empirical analysis shows that the accuracy and yield of the model in quantitative trading are superior to the traditional methods, and the high-yield and low-risk investment objects selected by DEA significantly improve the robustness of the strategy, and the feasibility and superiority of the system in financial investment decision-making are verified by the application of the power sector. Therefore, the research in this paper not only optimizes the quantitative trading model, but also provides theoretical support and practical basis for risk management and intelligent investment strategies in financial markets.