應用資料探勘於自動化交易策略之獲利能力評估的模型與平台設計及建置;The Design and Implementation of a Platform for Analysis the Risk and Profit of Trading Strategies via Data Mining Technology.

NCUIR > School of Management at National Central University > Graduate Institute of Information Management > Electronic Thesis & Dissertation > Item 987654321/74757

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/74757

Title:	應用資料探勘於自動化交易策略之獲利能力評估的模型與平台設計及建置;The Design and Implementation of a Platform for Analysis the Risk and Profit of Trading Strategies via Data Mining Technology.
Authors:	顏茂城;Yan, Mao-Cheng
Contributors:	資訊管理學系
Keywords:	分類型參數;連續型參數;資料探勘;視覺化圖表;自動化平台;Classification Parameters;Continuous Parameters;Data Mining;Visualization Chart;Automation Platform
Date:	2017-07-05
Issue Date:	2017-10-27 14:38:26 (UTC+8)
Publisher:	國立中央大學
Abstract:	本研究選用布林通道與凱特通道，結合濾網與出場機制，依據分類型參數與連續型參數概念設計多個子策略。接著，選取台灣50、台灣高股息、中型100、治理100、上市股票、上櫃股票，針對6個資料集執行多個子策略的最佳化回測，最佳化回測後將產生巨量績效資料，再利用高原自動搜尋演算法，尋找淨利績效高原。尋找高原後，本研究進一步做完整性分析，根據子策略的連續型參數特性，以多維度高原圖呈現，提供子策略內部評估；根據子策略分類型參數特性，以各公司熱區圖呈現，提供子策略與子策略外部評估；根據資料集多股票的比較特性，以幾何圖形之互動式平行座標圖呈現，提供股票與股票整體性評估。另外，本研究亦將核心流程與功能串聯、整合，並建置至 Python Django自動化平台，藉由自動化的優點，主動運算績效結果、主動進行巨量績效資料分析、最後主動提供視覺化圖表，使投資者根據視覺化圖表就可透視巨量績效資料所帶來的隱含相關性與意義，協助投資者選用優等的股票與子策略。 ;This research used Bollinger channel and Keltner channel, combined with a filter net and exiting mechanism, according to the classification parameters and continuous parameters concept design of multiple sub-strategies. Then, choosing Taiwan 50 ETF, Taiwan dividend plus ETF, Taiwan mid-cap 100 ETF, TWSE corporate governance 100 ETF, the stock of listed company and, the stock of listed company in OTC, for the six data sets to implement multiple sub-strategies optimization, after optimizing will produce a huge amount performance information of sub-strategies, and further using the algorithm of plateau automatic searching to look for high net profit performance of plateau. After looking for the high performance plateau, according to the continuous parameters characteristics of the sub-strategies, provided multidimensional plateau visualization chart to evaluate the internal relation of the sub-strategies. According to classification parameters characteristics of the sub-strategies, provided heatmap visualization chart of companies to evaluate the external relation of the sub-strategies. According to the comparative characteristics of the stock, provided the interactive parallel coordinate visualization chart of geometric figures to do stock and stock comprehensive evaluation. In addition, this research also integraed the core process and functions, and built into the Python Django automation platform, through automation of the advantages, initiative on computing performance results, initiative on analyzing a huge of performance data, finally initiative on providing visualization charts, let investor can see the implicit relevance and significance of huge amounts of performance data by visualization charts to help investors use superior stock and sub-strategies.
Appears in Collections:	[Graduate Institute of Information Management] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	211	View/Open

社群 sharing

Loading...