Article Preview
TopLiterature Review
Deep learning, as an important tool for stock market prediction and investment decision-making research in the current financial management field, has been explored by many scholars for the application of deep learning algorithm models in stock market prediction and investment decision-making. In stock market prediction in financial management, Nabipour et al. (2020) analyzed historical stock price data and predicted future trends by using Convolutional Neural Networks (CNN) and LSTMs. The research results showed that the combination of the two deep learning algorithm models outperformed other time series-based methods in terms of accuracy in stock market prediction. The research of Rouf et al. (2021) has shown that deep learning models perform well in terms of accuracy and stability in stock market forecasting and trading decisions. They used deep belief networks (DBN) and autoencoders (AE) for prediction and found that these models performed well in predicting market fluctuations and oscillations. At the same time, the accuracy of feature extraction is also crucial for the predictive performance of the model (Rouf et al. 2021). Therefore, some researchers use convolutional and pooling layers in CNN to capture the characteristics of time series data such as stock prices and trading volumes, thereby improving the accuracy of stock trend prediction (Lee et al., 2021). Feng et al. (2022) used adaptive compressive sensing (CS) technology for feature extraction and modeling of BILSTM models, and they successfully predicted the trend of the Shanghai and Shenzhen 300 Index.
In addition, by analyzing and predicting the trends of the stock market in financial management, investors can better understand market trends and possible development directions, making wiser investment decisions, guiding enterprises in investment, financing, and capital structure optimization decisions, and helping enterprises achieve more stable profits and risk management. However, when making investment decisions, investors need to consider multiple factors comprehensively and conduct in-depth analysis of relevant investment decisions in order to improve the efficiency and returns of stock market investments (Fan & Peng, 2022). Singh et al. (2022) used a combination model with feedforward (BP) neural network as the core technology to predict the return of Shanghai Stock Exchange and achieved good prediction results. In recent years, scholars have studied the use of reinforcement learning models to formulate investment decision strategies. Through interaction with the environment, the model gradually learns and improves strategies (Buczynski et al., 2021). Some researchers have used reinforcement learning models for stock investment decisions and achieved good results.
There is still no perfect algorithm model capable of accurate prediction of the trends and investment decisions in the stock market in financial management. The volatility and uncertainty of the stock market make it difficult to achieve precise predictions, and a model may exhibit errors in the forecasted outcomes. Based on the above research and deep learning technology, this paper proposes a new RCA-BiLSTM-DQN stock intelligent prediction and investment decision-making method. It integrates MTS stocks to represent multi-temporal features and employs RCA for feature extraction to capture market dynamics and trend information comprehensively. Then, the IWOA intelligent optimization algorithm is used to train BiLSTM and determine optimal parameters. The prediction results are then combined with the DQN network investment strategy model to construct a self-adaptive and universal RCA-BiLSTM-DQN stock intelligent model. This proposed model contributes to an improvement in prediction accuracy and informed investment decisions, yielding greater benefits.