Abstract: In recent years, there are some complex typical characteristics of financial market, such as structural break and risk contagion. These characteristics may bring great challenges to risk management and investment portfolio, and structural break in financial market is the key to financial risk contagion, whereas risk contagion is a tough problem that should be solved urgently in the research of investment portfolio. In this paper, the exchange rate of 9 major international currencies against the US dollar is taken as the research object. Firstly, Hidden Markov Model (HMM) is constructed to predict the structure break of exchange rate assets according to the characteristics of structural break that may occur in the exchange rate market, and then dynamic R-Vine Copula model is constructed to predict the risk contagion relationship of exchange rate assets, combined with the the risk contagion relationship that may be induced by structural break between exchange rate assets. Finally, based on the research results of structure break and risk contagion, portfolio exchange rate assets screened out, and a time-varying portfolio forecasting model constructed. The empirical results showed that: HMM model can effectively depict the structure break in exchange rate assets; dynamic R-Vine Copula model can more effectively depict the risk contagion relationship between exchange rate assets; dynamic portfolio forecasting model based on the relationship between structure break and risk contagion has the advantages of reducing portfolio risk and improving portfolio forecasting return.Therefore, an operable research scheme is provided for structure break and risk contagion in financial market, risk supervision departments and institutional investors.

Key words: structure break; risk contagion; high-dimensions exchange rate assets; dynamic portfolio forecasting