- 學位論文
結合選股與支援向量迴歸解決穩健投資組合問題
Combining Stock Selection and Support Vector Regression to Solve Robust Portfolio Problem
摘要
金融海嘯後，各國央行挹注資金以挽救金融市場及投資人信心，台灣央行在2009年2月19日宣佈降息來到1.25％，為歷史上的新低點。投資人將資金採用定存在銀行只有極低的利率，同時還得面臨通貨膨脹之風險，使得投資人紛紛將資金轉移至金融市場進行投資。股票市場進入門檻較低，且較為投資大眾熟悉，所以大多數的投資者皆選擇股市來進行投資，故股市投資組合問題實為一重要議題。而在金融海嘯過後，投資人在投資策略上傾向於保守與穩健，本研究則欲提供投資人穩健的投資組合組成方式。 本研究欲從資本資產訂價模型出發，以該模型中解釋個別資產報酬率與市場報酬率之貝他係數，先挑選出與市場報酬率連動性相對高的產業，並結合特殊公司特性來挑選各產業下較有可能為投資帶來超額報酬之個股。隨後以時間序列結合支援向量迴歸對個股之可能報酬率與風險值進行預測，最後以預測出之期望報酬率與風險值來建構出投資組合基本模型，並於預測之期望報酬部分導入穩健的觀念，最後求解出穩健資產配置權重。 實證結果證明本研究所求出之穩健投資組合在整個大盤處於表現不佳時，能有效的降低投資人的損失，同時延伸出之混合選股策略，比起僅使用單一選股策略更能進一步降低投資人的損失。
並列摘要
After the financial tsunami, central bank of nations inject money to save financial markets and investor confidence, the central bank of Taiwan announced the interest rates cuts to 1.25% in February 19, 2009, it’s the lowest record in history. Investors who save funds in the bank not only just can obtain a very low interest rates,but also face the risk of inflation, so investors have to transfer funds to invest in financial markets. Stock market is easy to entry, and more familiar with the investors, so that most of investors are choosing to invest in stock market, stock portfolio problem is really a important issue. After the financial tsunami, investors tend to be conservative and robust in investment strategy, this research want to provide investors a robust portfolio composition method. This study starts with the capital asset pricing model, first, choose the industry which is highly related to market rate of return by using the beta coefficient of capital asset pricing model which to explain individual asset returns and market returns, and combined with special features of corporation to selected stocks and these stocks may bring excess return for investment. Subsequently combined with the time series and support vector regression to predict the possible rates of return and value at risk of individual stocks, and then using predictive expected rates of return and value at risk to construct the portfolio basic model, simultaneously, at the part of predictive values imports the concept of robust, finally solving the robust asset allocation weights. The empirical results proved the robust portfolio can effectively reduce the loss of investors when market is depression, simultaneously this study extend a mixture of stock picking strategy, compared to single stock selection strategy, the mix strategy can reducing more loss of investors.