以邏輯迴歸模型建立之戰略性資產配置擇時交易策略

Abstract

戰略性資產配置(Tactical Asset Allocation)有別於傳統的買進持有策略(Buy and Hold)，透過特定的判別機制(如總濟、個經變數或特殊事件)，積極地在投資組合的資產間轉換，以尋求優於單一資產的報酬。本研究延續Arshanapalli, Switzer, Hung[2004]所提出的TAA邏輯迴歸模型，並在交易決策時點加入移動平均線(Moving Average)修正概念，成功地證明擇時策略(Market Timing)在台灣市場的可行性。回溯測試的結果更顯示：加入MA 指標後，模型的預測能力得以全面提升，以[股票 vs. 長債]及[股票 vs. 現金]為資產配置的投資組合分別可獲得24.67%及19.86%的顯著年化報酬，優於投資組合中任何單一資產買進持有的結果。測試期間涵蓋台股多頭及空頭時期，在考慮交易成本的情況下，本文建立了一個以公開資訊為基礎的投資預測模型，並可廣泛應用於基金經理人及一般投資大眾。

Tactical Asset Allocation, TAA, is evidence on the predictability of stock returns using macroeconomic, microeconomic and event tactical variables. Portfolio managers are seeking higher-than-single-asset revenue (Buy and Hold) by actively investing between assets. This article extends the Arshanapalli, Switzer, Hung[2004]’s TAA logistic regression model by adding a Moving Average criteria. The MA model not only enhanced the performance of ASH Model, but also has proven the feasibility of the Market Timing strategy in Taiwan capital market. Back-tested results of [Stock vs. Bond] and [Stock vs. Cash] portfolios exhibited significant annualized returns of 24.67% and 19.86%, respectively. Gives consideration of transaction costs and holding period, this strategy which earns excess return than any single asset is inconsistent of the efficient market hypothesis. In this article, we have successfully established a predictive model based on public information, and can be applied widely by fund managers and individual investors.