In this study, we construct an optimal currency portfolio using the implied return distribution in the mean-variance approach and examine the performance through a backtest. We estimate the implied expected spot return, implied volatility, and implied correlation from currency option price data, and propose a method of constructing a fully forward-looking optimal currency portfolio without historical data. We implement the backtest from January 2006 to October 2020 on a currency portfolio comprising seven currencies (the Japanese yen, the Swiss franc, the euro, the British pound, the Australian dollar, the New Zealand dollar, and the Canadian dollar) against the US dollar and US-dollar interest rate, and examine the usefulness of the proposed method. We find that the proposed method yields a higher performance than the conventional method in previous studies that use historical data. Furthermore, it is evidenced that the main factor in the performance gap between the proposed and the conventional methods is the high predictive power of the spot return.

在本研究中，我们使用均值方差方法中的隐含收益分布构建最优货币投资组合，并通过回测检查其表现。我们从货币期权价格数据中估计了隐含预期即期收益、隐含波动率和隐含相关性，并提出了一种在没有历史数据的情况下构建完全前瞻性的最优货币投资组合的方法。我们对2006年1月至2020年10月7种货币（日元、瑞士法郎、欧元、英镑、澳元、新西兰元、加元）对美元的货币组合进行回测美元和美元利率，并检验所提出方法的有用性。我们发现所提出的方法比先前使用历史数据的研究中的传统方法具有更高的性能。此外，事实证明，所提出的方法与传统方法之间性能差距的主要因素是现货回报的高预测能力。