利用時間序列預測ETF─以台灣50為例

Using Time Series to Forecast ETF─Taiwan 50 as an Example

鄧勝昌(Sheng-Chang Teng)

指導教授 ： 邱南星 副教授

健行科技大學/商管學院/資訊管理系碩士班/碩士(2018年)

摘要

本研究是運用時間序列做為預測台灣50 ETF（Exchange Traded Funds）股票指數的研究，近年來通貨膨脹率一直提高，利用存款方式將財富存到銀行的利息太低。因此，投資理財將是累積財富重要課題。在投資方式中，買賣股票最容易入門。只要到證券公司開戶，再準備一個銀行帳戶，透過網路或電話就可下單。在各種金融商品中，股市的資訊是公開且容易取得，但投資股市的風險較定存或公債等投資為高。若能有效掌握漲跌趨勢，就可避開風險而得到豐厚的報酬。
市場上有許多投資者透過各種的分析與預測模型，希望能準確地預測股價。若股價能被有效準確地預測，就能在股市中除了能規避風險而且能賺取差價報酬。一般有許多不同的方法預測股票價格的方法。一般投資人最普遍的方法有基本分析、技術分析、籌碼分析與消息分析。本研究是幫助無法長時間看盤及沒學習技能分析的投資者。經實驗結果為使用交易日到交易日前46日的區間歷史資料及配多支持向量機演算法，去預測隔天交易日收盤價來決策買賣點最精準的。

關鍵字

ETF ； 時間序列 ； 股票預測

並列摘要

This study uses time series as a predictor of Taiwan 50 ETF (Exchange traded funds) stocks, which have been rising in recent years and have been too low on deposits to deposit money with banks.Therefore. In the way of investment, buying and selling stocks is easiest to get started. In a variety of financial commodities, the stock market information is open and easy to obtain, but the risk of investment in the stock market is higher than the fixed deposit or public debt investment.If you can effectively master the trend of ups and downs, you can avoid the risks and get rich rewards.
There are many investors in the market through various models of analysis and prediction, hoping to accurately predict the stock price. If the stock price can be effectively and accurately predicted, it can be in the stock market in addition to avoid risk and can earn the difference pay. There are generally many different ways to predict stock prices.The most common methods of general investors are basic analysis, technical analysis, chip analysis and message analysis.This study is to help investors who are unable to look at the disk for a long time and fail to learn the skill analysis. The experimental results for the use of the trading day to 46 days before the date of the interval historical data and support vector machine algorithm, to predict the next day trading closing price to decide the most accurate trading point.

並列關鍵字

ETF ； Time Series ； Stock Foreca

參考文獻

[1] 王威能 (2015)。「依據技術指標投資台灣 50 ETF 及其成分股」。國立台北大學經濟學研究所，碩士論文。

[2] 余重毅 (2014)。「移動平均線在台灣股票指數期貨的應用」。逢甲大學財務金融學系，碩士論文。

[3] 李怡恬 (2004)。「效率市場假說之實證研究－以亞太地區股市為例」。銘傳大學，碩士論文。

[4] 李彥撰 (2011)。「以高頻率日內資料探察台灣50ETF價格發現功能」。德明財經科技大學，碩士論文。

[5] 邱一薰 (2005)。「類神經網路預測台灣 50 股價指數之研究」。國立彰化師範大學資訊管理研究所，碩士論文。