2021 / September

新聞情緒指標與臺灣加權股價指數之關係
Relationship between News Sentiment Indicator and the Taiwan Weighted Stock Index

研究論文 Research Article

陳冠臻, 林忠億, 陳宏銘
Kuan-chen Chen, Chung-i Lin, Hong-ming Chen

383-423

本研究以媒體文字量化資訊作為新聞情緒（news sentiment）之變數，探討新聞情緒與臺灣加權股價指數報酬率的關係。結果發現正向、負向情緒變數與股價指數報酬率具有統計上的顯著性結果。我們進一步建立投資策略並考量交易成本於做多的績效，發現加入負向情緒變數後之投資策略，其最佳收益率為13.308%、夏普指標為 0.759、期末現值為 197.528（萬），皆比未納入情緒指標 來得高，顯示納入負向新聞情緒於績效指標均表現較好。本研究結果顯示以新聞正、負向內容形成的新聞情緒指標的確能預測股票市場報酬率，且負面情緒 預測能力較正面情緒來得好，與 Tetlock（2007）及 García（2013）的研究結果一致。   

This study explores the relationship between positive and negative news sentiment and the returns on the Taiwan weighted stock index by taking media quantitative information as a news sentiment variable. The findings show that these two sentiment variables and the returns on the stock index are statistically significant. Through further establishment of investment strategies and valuing a long position on an option with transactions costs, this study shows that, for the investment strategy with the negative sentiment, the optimal return is 13.308%, the Sharpe ratio is 0.759, and the ending present value is 1,975,280 thousand. They are higher than those strategies with no sentiment variable. This denotes that the performance indicator is better if the negative sentiment variable is included. The results of this study indicate that the negative sentiment is a better prediction than is positive sentiment, which is consistent with the results of Tetlock (2007) and García (2013).

新聞情緒, 投資人情緒, 股價指數, 文字探勘, 交易策略
news sentiment, investor sentiment, stock market index, text mining, trading strategy

經濟學
Economics

10.53106/1018189X2021093303001

Abstract Full Text