台灣股票市場中,散戶為重要的投資者之一,根據行政院金管會的證期局資料統計,在2011年台灣自然人成交金額佔總市場成交金額62.7%,佔所有投資人類別中投資金額比例最高,在國內散戶參與度頗高的情況下,投資行為就受到許多關注。根據過去文獻中發現,投資人情緒是會影響投資決策,而投資決策影響股價表現,所以投資人情緒與股價是有其相關性的。 本研究蒐集股市討論區文章,並透過文字探勘技術建構投資人情緒指標,以建立台灣加權股價指數趨勢預測模型,實驗結果發現,結合情緒指標及股市相關指標(國際股價指數、總體經濟指標)資料所建立的預測模型,在上漲的預測能力,達到65.63%的正確率,下跌的預測能力其正確率達到70%,而整體的正確率有67.24%,實驗模型中結合兩種資訊能有效提升模型的預測能力。
Retail investor is one of the important investors in the stock market of Taiwan. According to the FSC Securities and Futures Bureau statistics, the total market turnover create by natural person is 62.7% in 2011,Which is accounts for the highest proportion of investment amount in all categories of investors. In the domestic retail investor participation in the high case, the investment behavior received much attention. Based on past literature, investor sentiment will affect investment decisions, and investment decisions affect the stock price performance, So we got to the conclusion that investor sentiment and stock price are related.In this study, we collection stock message board article, through text mining to construct the sentiment indicators to point the trend of taiwan stock exchange capitalization weighted stock index. The experiment shows that our model combines the sentiment indicators and related indicators of stock , it's able to determine the predictive ability of rise to 65.63% accuracy, the predictive ability of fall to 70% accuracy and the overall may achieve 67.74% accuracy. The experiment model combines above two kinds of information can effect to enhance the predictive capability.
為了持續優化網站功能與使用者體驗,本網站將Cookies分析技術用於網站營運、分析和個人化服務之目的。
若您繼續瀏覽本網站,即表示您同意本網站使用Cookies。