Research on China's Stock Index Futures Pair Trading Strategy Based on High-Frequency Data

The efficiency and accuracy of high-frequency trading strategies are increasingly emphasized in modern financial markets, especially in China's stock index futures market. Bayesian theory and probabilistic neural network algorithms have become important tools for constructing high-frequency trading models because of their powerful predictive ability and adaptivity. By studying the minimum error rate and minimum risk Bayesian decision in Bayesian theory, as well as the structure and training methods of probabilistic neural networks, this thesis aims to develop a pair trading strategy for stock index futures based on high-frequency data. The division of dataset and the classification method of classically unbalanced dataset are the key steps of the research, through which the performance of the model is optimized and the effectiveness of the strategy is verified through back-testing experiments. This study not only improves the accuracy and stability of the high-frequency trading strategy, but also provides a new idea and method for quantitative trading in China's financial market.