双融合模型在香港股市的波动率量化交易策略研究

Research on Double Fusion Modeling for Volatility Quantitative Trading Strategies in Hong Kong Stock Market

摘要: 在金融市场中，波动率反映了资产价格的不稳定性和风险程度。波动率量化交易策略是一种利用波动率来指引投资者进行交易的方法，并能够带来一定的收益。本文的目的是探讨时间序列模型和机器学习模型相结合的双融合模型在波动率量化交易策略中的应用，以提升策略的效果和稳健性。我们采用了3种时间序列模型Generalized Autoregressive Conditional Heteroskedasticity (GARCH)、Glosten Jagannathan Runkle GARCH (GJR-GARCH)和Exponential GARCH (EGARCH)，以及4种传统机器学习模型Random Forest (RF)、Adaboost (ADA)、Gradient Boosting Decision Tree (GBDT)和Histogram Based Gradient Boosting (Hist-GB)，组合成了12种双融合模型对波动率进行预测。同时，我们进行了模拟交易的回测实验，以评估该波动率量化交易策略的性能。该策略在香港市场中的股票都获得了较好的收益，为投资者提供了有价值的参考。

Abstract: In the financial markets, volatility is indicative of the fluctuation in asset prices and the associated level of risk. The volatility quantitative trading strategy is an approach that leverages volatility to make trading decisions and generate specific returns. This paper aims to explore the application of double fusion modeling that combines time series models and machine learning models within the volatility quantitative trading strategy to augment its efficiency and stability. We employ 3 time series models: Generalized Autoregressive Conditional Heteroskedasticity (GARCH), Glosten Jagannathan Runkle GARCH (GJR-GARCH), and Exponential GARCH (EGARCH); alongside 4 traditional machine learning models: Random Forest (RF), Adaboost (ADA), Gradient Boosting Decision Tree (GBDT), and HistogramBased Gradient Boosting (Hist-GB). These models are amalgamated into 12 double fusion models for prediction analysis. Meanwhile, we execute backtesting trading experiments to assess the strategy’s performance. The strategy has demonstrated superior returns in Hong Kong market, and offers a valuable trading reference for investors.