投資股票市場是一種滿熱門的投資管道，但卻存在著許多影響股價的各種因素。在資訊科技不斷發展之下，已有許多學者嘗試以各種理論與方法來預測股票市場的未來走勢，其中以人工智慧方法最為引入注目。 支援向量機（Support Vector machine, SVM）是一種基於結構風險最小化的新穎學習機器演算法。由於它具有許多吸引人的特性以及可靠的經驗效能，因此有越來越多學者將它應用到各個領域中。 本研究嘗試以支援向量機為分析工具，運用於台灣股票市場。以股市加權指數為樣本，利用嵌入維度切割股價指數報酬序列，並透過支援向量機來建構分析模型，產生股價指數之漲跌預測正確率，同時呈現訓練資料多寡對其影響。 研究結果顯示，在一般的情況下，台灣股票市場指數波動是隨機的。本研究得知，在訓練資料數量方面，長期的訓練資料並無明顯注於未來預測；反之，中短期數量的訓練資料較能抓住未來走勢。 Stock market is a popular financial tool. There are many factors that affect stock prices. Computer technology has been constantly improving, and there is a growing interest in applying more sophisticated mathematical tools in studying the stock market. Support Vector Machine (SVM) based on structural risk minimization theory is a modern algorithm of learning machine. Lots of scholars apply SVM to different kinds of problems due to many attractive features and promising empirical performance. In this paper, we try to use SVM as an analytical tool and apply it for analyzing Taiwan stock market. Return series constructed from The Taiwan Stock Exchange Capitalization Weighted Stock Index (TAIEX) time series and cutting by embedded dimension are regarded as the original data. Support vector machine is applied to construct analytical model for the stock index fluctuation simulation. Results reveal that the fluctuation of TAIEX is random walk in general. In amount of training history data, it shows that long training data period is not strikingly helpful to predict the trend of the stock index, but using medium-term or short-term training data is good for catching the future stock index''s tendency.
