標題: 以1分單位建構9與30分輪廓圖極短線交易行為比較分析
Creating 9 and 30 Timeframes of Market Profiles for Trading Behavior Analysis Based on 1-minute intervals of TPO
作者: 洪敏智
Hung, Min-Chih
陳安斌
Chen, An-Pin
管理學院資訊管理學程
關鍵字: 類神經網路;市場輪廓指標;技術分析指標;臺灣指數期貨;Artificial Neural Network;Market Profile;Technical Analysis;Taiwan Stock Index Futures
公開日期: 2012
摘要: 在市場輪廓(Market Profile)理論中,其最短期的結構是通過市場輪廓圖(TPO Chart)將每30分鐘的交易價格反映出來,當進行市場交易時,市場開始形成當日的交易特性。當一天的交易日結束時,市場輪廓圖(TPO Chart)描繪了市場參與者的行為。現行交易市場揭露「指數」頻率為每15秒1次,但以30分為單位TPO單位建構之300分鐘市場輪廓圖,用以解釋市場行為明顯不夠精致敏感。 本研究採用類神經網路為模型建構之人工智慧系統,配合輸入使用1分鐘TPO建構而成的9及30分鐘市場輪廓圖指標之物理力量,進行動態時空環境下的自我學習,預測未來5、15及30分鐘趨勢方向,發掘台灣指數期貨市場趨勢行為知識。 由實驗結果得知使用市場輪廓指標在極短線預測(時間為5分鐘)的趨勢方向判定優於技術分析指標,且增加市場輪廓歷史資訊對極短線交易的趨勢方向判斷上有顯著的幫助,證明以1分為單位建構的市場輪廓指標的在極短線市場具有良好的預測未來趨勢能力,可提供投資人更明確的買賣交易資訊,以輔助決策者做正確抉擇的依據。
Market Profile, the most short-term market structure contour map by TPO every half hour of trading will be reflected in the price, when the market transactions, the market began to form the day of transaction characteristics. When the day is the end of the trading day, the Market Profile depicts the behavior of market participants. Taiwan Stock Exchange (TWSE) has increased the frequency of statistical information disclosure on TSEC weighted index from originally 1 minute to 15 seconds, but every half hour TPO market units Market Profile of 300 minutes, the market behavior is not enough to explain the exquisite sensitivity. In this study, neural network modeling of artificial intelligence for the system, with input TPO Chart using 1-minute intervals of TPO to Creating 9 and 30 Timeframes of Market Profile physical forces, space-time environment for dynamic self-learning, to predict the future 5, 15 and 30 minutes trend direction, identify trends in behavior of the Taiwan index futures market knowledge. The experiments demonstrate Market Profile outperforms Technical Analysis model in forecasting accuracy of a short-term time interval, providing better advice on investment operation.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT070063426
http://hdl.handle.net/11536/71818
顯示於類別:畢業論文