Straddle Option is a financial trading tool that explores volatility premiums in high-volatility markets without predicting price direction. Although deep reinforcement learning has emerged as a powerful approach to trading automation in financial markets, existing work mostly focused on predicting price trends and making trading decisions by combining multidimensional datasets like blogs and videos, which led to high computational costs and unstable performance in high-volatility markets. To tackle this challenge, we develop automated straddle option trading based on reinforcement learning and attention mechanisms to handle unpredictability in high-volatility markets. Firstly, we leverage the attention mechanisms in Transformer DDQN through both self-attention with time series data and channel attention with multi-cycle information. Secondly, a novel reward function considering excess earnings is designed to focus on long-term profits and neglect short-term losses over a stop line. Thirdly, we identify the resistance levels to provide reference information when great uncertainty in price movements occurs with intensified battle between the buyers and sellers. Through extensive experiments on the Chinese stock, Brent crude oil, and Bitcoin markets, our attention-based Transformer-DDQN model exhibits the lowest maximum drawdown across all markets, and outperforms other models by 92.5% in terms of the average return excluding the crude oil market due to relatively low fluctuation.

This paper proposed an innovative WTC-LSTM-ASSA-NSTformer framework for gold price forecasting. The model integrates Wavelet Transform Convolution, Long Short-Term Memory networks (LSTM), and an improved Nyström Spatial-Temporal Transformer (NSTformer) based on Adaptive Sparse Self-Attention (ASSA), effectively capturing the multi-scale features and long- and short-term dependencies of gold prices. Additionally, for the first time, various financial and economic uncertainty indices (including VIX, GPR, EPU, and T10Y3M) are innovatively incorporated into the forecasting model, enhancing its adaptability to complex market environments. An empirical analysis based on a large-scale daily dataset from 1990 to 2024 shows that the model significantly outperforms traditional methods and standalone deep learning models in terms of MSE and MAE metrics. The model’s superiority and stability are further validated through multiple robustness tests, including varying sliding window sizes, adjusting dataset proportions, and experiments with different forecasting horizons. This study not only provides a highly accurate tool for gold price forecasting but also offers a novel methodological pattern to financial time series analysis, with important practical implications for investment decision-making, risk management, and policy formulation.

This study introduces a novel factor model based on self-attention mechanisms. This model effectively captures the non-linearity, heterogeneity, and interconnection between stocks inherent in cross-sectional pricing problems. The empirical results from the Chinese stock market reveal compelling ffndings, surpassing other benchmarks in terms of profftability and prediction accuracy measures, including average return, Sharpe ratio, and out-of-sample R2. Moreover, this model demonstrates both practical applicability and robustness. These results provide valuable evidence supporting the existence of the three aforementioned properties in crosssectional pricing problems from a theoretical standpoint, and this model offers a powerful tool for implementing profftable long-short strategies.

Deep learning technology is rapidly adopted in financial market settings. Using a large data set from the Chinese stock market, we propose a return-risk trade-off strategy via a new transformer model. The empirical findings show that these updates, such as the self-attention mechanism in technology, can improve the use of time-series information related to returns and volatility, increase predictability, and capture more economic gains than other nonlinear models, such as LSTM. Our model employs Shapley additive explanations (SHAP) to measure the “economic feature importance” and tabulates the different important features in the prediction process. Finally, we document several economic explanations for the TF model. This paper sheds light on the burgeoning field on asset allocation in the age of big data.