在這項研究中，我們著重研究在定期的重新分配之前，先檢驗重新平衡的必要性。在product partition model假設下，通過changepoint detection來進行檢測。 我們提出一種新的重分配策略dynamic rebalancing strategy with optimal training period（DRO）進行動態的重新平衡，以改善定期重新分配策略。我們通過回測測試來檢驗我們的DRO策略的效果並與定期重新分配策略的成果進行比較。 最後，我們發現DRO策略在較長的時間間隔假設下，復合年增長率(CAGR)方面有更大的回報。另外，當經濟形勢穩定時，DRO策略的表現優於 定期重新分配策略。 ;Reallocation, or adjust weights of portfolio is an indispensable part in portfolio management. In the practice, calendar rebalancing is a basic rebalancing strategy that either retail or institutional investors can utilize to create an optimal investment process. In calendar rebalancing, portfolio managers reallocate their portfolio at predefined intervals and use the historical data over the pass fixed time to calculate the suitable weights. It′s known that each time you rebalance the portfolio, paying for the tax and transaction fee is inevitable.However, reallocating the portfolio does not always get the relevant return.
In this study, we focus on examining the necessity of rebalancing before the regular reallocation by using changepoint detection under a product partition model. We propose a dynamic rebalancing with optimal training period (DRO) to improve the calendar rebalancing. We examine the efficiency of our rebalancing strategy by using backtesting procedure and compare with the calendar rebalancing. As a result, we discover that the DRO strategy has greater reward in terms of compound annual growth rate when the rolling window is longer. Besides, the representation of the DRO strategy is better than the calendar rebalancing in general when the economic situation is steady.
