熊市預測投資策略之穩健性分析

Abstract

本文旨在檢驗馬可夫轉換模型(Markov-switching model)在市場預測上的有效性，我們使用馬可夫轉換模型辨認過去期市場趨勢為熊市(bear market)或牛市(bull market)，然後透過數種總體經濟變數或者財務變數來預測下期市場狀態並建立交易策略。此外，本文亦嘗試比較透過不同方法－以預測得的下期報酬或者下期熊市機率－所建構出的交易策略之間的優劣。根據我們實際執行樣本外檢測(out-of-sample test)所得到的結果，以預測報酬所建立的策略有較佳的表現。而關於市場預測的有效性，總的來說，以馬可夫轉換模型所建立的交易策略確實能帶來超額報酬，而其中又以使用長短期利差(term spread)、通貨膨脹率(inflation rate)、違約殖利率差(default yield rate)、聯邦基金利率變動率(change in the federal funds rate)、貨幣供給增長率(money growth)以及股市變異數(stock variance)所建立的交易策略成效較佳。

This study examines the usefulness of the Markov-switching model for its economic profitability. We predict the recessions in the stock market by various macroeconomic and financial variables and build the trading strategies accordingly. This study also focuses on the comparison between the strategies by criteria of predicted return or bull market probabilities. Empirical result of the out-of sample experiments shows that the strategies conducted with predicted return perform better than those based on probabilities. About the usefulness, overall, the trading strategies based on the Markov model and our predictors, especially term spread, inflation rate, default yield rate, change in the federal funds rate, money growth and stock variance, create excess return even on a risk-adjusted basis which proves the profitability of the market timing strategies.