Dynamic allocations for currency investment strategies
本文对五种货币投资策略进行样本外测试，发现基于预测模型的动态配置策略在统计和经济指标上均优于使用历史均值的基准模型，年化经济收益约1.16%。
This study conducts out-of-sample tests for returns on individual currency investment strategies and the weights on the universe of these strategies. We focus on five investment strategies: carry, momentum, value, dollar carry, and conditional FX correlation risk. The performances of our predictive models are evaluated using both statistical and economic measures. Within a dynamic asset allocation framework, an investor adjusts investment strategy weights based on the results of the prediction models. We find that our predictive model outperforms our benchmark, which uses historical average information in terms of statistical and economic measures. When the Sharpe ratio of the benchmark model is 0.52, our predictive model generates an economic gain of approximately 1.16% per annum over the benchmark. These findings are robust to the changes in investors’ risk aversion and target volatility for portfolio optimization.