Abstract: As financial time series have the characteristics of dynamic time-varying, asymmetric and nonlinear correlation, the article presented a mixture Copula model which had time-varying weight coefficients and Copula parameters. Meanwhile, based on mixture Copula models whose weight coefficients and Copula parameters were both time-varying, to describe the correlation between different frequency of internet banking stocks price series, then we construct a new kind of pairs trading strategy model, and compared with that of static mixture Copula models. Empirical analysis shows that pairs trading strategy based on time-varying mixture Copula models gains a high stable returns. Both of weight coefficients and Copula parameters are time-varying can capture more trade opportunities and strategy performance is the best. Mixture Copula models which have more Copula functions don’t have the advantage in pairs trading strategy. High frequency financial markets are more profitable than the corresponding low frequency financial markets.

摘要： 由于金融时间序列数据具有动态时变、非对称和非线性相关的特征，本文给出了一类权重系数和Copula参数均时变的混合Copula模型。同时基于权重系数和Copula参数时变的混合Copula模型，刻画不同频率互联网金融概念股价格序列之间的相关性，构建配对交易策略模型，并与静态混合Copula模型下的策略结果进行比较。实证分析表明：基于时变混合Copula模型的配对交易策略可以获得较高收益；Copula参数和权重系数均时变的混合Copula模型能捕获更多交易机会，策略表现最好；含有较多成分Copula的混合模型在配对交易策略中并不具有优势；高频率金融市场比相应低频率金融市场的策略盈利更高。