在股票市場上許多投資人致力於股價的預測,但股票價格常常受到許多人為因素、政治因素、經濟因素、突發事件或是其他未知因素的影響,使得建立一個準確的預測模型相當不易。本研究利用時間序列法進行股價預測,並以平均絕對百分比誤差 (mean absolute percent error;MAPE) 進行預測的效能分析。然而,雖然利用時間序列法的預測誤差較小,但對於股價漲跌的預測準確度卻不高。為此,本研究以倒傳遞類神經網路 (Back Propagation Neural Network;BPNN) 為基礎,結合多個技術指標進行股價漲跌的預測。最後,本研究針對「台中銀行」、「台積電」、「大立光」、「鴻海」和「力旺」五家公司的股價資料進行實證分析。研究結果顯示,倒傳遞類神經網路對於股價漲跌的預測正確率有不錯的效果。
Many investors in the stock market are committed to predict the stock price. However, the stock price could be affected by many artificial factors, political factors, economic factors, sudden accidents or other unknown factors, making it difficult to establish an accurate prediction model. At first, the time series method is used to forecast the stock price in this study. The mean absolute percent error (MAPE) is applied to evaluate the prediction performance. Although the prediction error is smaller by using the time series method, the prediction result of the stock price fluctuation is not accurate enough. As a result, this study combines the Back Propagation Neural Network (BPNN) with multiple technical indicators to forecast stock price ups and downs. Finally, the empirical analysis is implemented in this paper by using the stock price data of Taichung Commercial Bank, Taiwan Semiconductor Manufacturing Company and Limited and Precision Co.,Ltd, Foxconn Technology Group and eMemory Technology Inc. The results show that the BPNN has a good effect on the prediction accuracy of stock price ups and downs.
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。