本論文利用常見的技術指標配合人工神經網路模型(artifical neural network,以下簡稱ANN) 來預測股價走勢分析。技術指標以兩種資料形式提供給ANN,分別是原始的指標指數以及將技術指標指數轉換成股價走勢的漲跌預測(0代表預測下跌,1代表預測上漲),然後再比較此兩種資料形式訓練出來的模型其預測股價走勢的準確度。本研究主要標的為台灣的股價走勢並以台積電的歷史股價資料作為研究資料。內容涵蓋近十年歷史股價,包含最高價、最低價、收盤價以及成交量。而研究結果顯示將技術指標指數先轉換成相對應的股價漲跌預測訓練出來的模型,相較直接用技術指標訓練出來的模型,準確度的確有些微的提高。
This thesis addresses problem of predicting direction of movement of stock with artificial neural network(ANN). Ten technical parameters are first computed using stock trading data (high, low, close prices, and volume). The ANN model is then trained by two sets of data. One is the raw numbers of these ten technical parameters and the other is representing these technical parameters as discrete up/down data. Taiwan stock market is the main focus and Taiwan Semiconductor Manufactory Company(TSMC) stock is used for study.Evaluation is carried out on 10 years of historical data from 2009 to 2019, include high, low, close prices, and volume. Experimental results show that the performance ANN model does improve when these technical parameters are represented as discrete up/down data.