DC 欄位 值 語言 DC.contributor 經濟學系 zh_TW DC.creator 張佑偵 zh_TW DC.creator Yu-Chen Chang en_US dc.date.accessioned 2021-6-30T07:39:07Z dc.date.available 2021-6-30T07:39:07Z dc.date.issued 2021 dc.identifier.uri http://ir.lib.ncu.edu.tw:444/thesis/view_etd.asp?URN=108429003 dc.contributor.department 經濟學系 zh_TW DC.description 國立中央大學 zh_TW DC.description National Central University en_US dc.description.abstract 本文以台灣股票市場的產業加權指數報酬作為預測標的，建構產業輪動投資組合，探討Lasso迴歸是否真的有助於投資決策，通過實證結果發現，樣本內結果顯示前一期產業報酬對於個別產業報酬有一定的預測能力，台灣各產業間存在連動性，且個別產業能預測其他產業以及被其他產業預測的預測能力不一。紡織纖維類、金融保險業、資訊服務業及建材營造類常被Lasso選擇預測其他產業的超額報酬率。此外，本文基於OLS post-Lasso方法去建構產業輪動投資組合，將預測出來的報酬率分五等分位，做多（做空）預測超額報酬率較好（較差）的產業加權指數，發現OLS post-Lasso投資組合的表現尚可，其年化平均超額報酬率和夏普比率均高於使用OLS來建構的投資組合，同時，OLS post-Lasso的投資組合在景氣較不好期間有較好的投資表現。 zh_TW dc.description.abstract This paper uses the industry-weighted index returns of Taiwan stock market as the target of forecast, constructs an industry-rotation portfolio, and explores whether Lasso regression is really helpful for investment decision-making. The in-sample results show that some industries can predict the excess return of other industries. The textile fiber, finance and insurance industry, information service industry, and building construction industry are often selected by Lasso to predict the excess return of other industries. In addition, this paper constructs an industry-rotation portfolio based on OLS post-Lasso, buys (sells) industry-weighted indexes that predicts highest (lowest) excess returns , and finds that the portfolio based on OLS post-Lasso has not bad performance, and its annual average excess return and sharpe ratio are both higher than the portfolio based on OLS. At the same time, the portfolio based on OLS post-Lasso has better investment performance during periods of bad economic conditions. en_US DC.subject Lasso zh_TW DC.subject OLS post-Lasso zh_TW DC.subject 預測迴歸模型 zh_TW DC.subject Lasso en_US DC.subject OLS post-Lasso en_US DC.subject Predictive regression en_US DC.title 產業依存與跨產業報酬預測性：機器學習方法之應用 zh_TW dc.language.iso zh-TW zh-TW DC.type 博碩士論文 zh_TW DC.type thesis en_US DC.publisher National Central University en_US