使用倒傳遞類神經結合加權移動平均法預估 台灣股市

Using Back Propagation Neural Network and Weighted Moving Average for Index Prediction in Taiwan Stock Market

劉建宏(Jian-Chien Liu)

指導教授 ： 林金城

大同大學/電機資訊學院/資訊工程學系所/碩士(2012年)

摘要

在股票市場當中，每天的大盤指數漲跌多少，單一個股股票的收盤、漲跌是大家關心的議題，常會遇到的是如何去評估隔天的指數多少、漲跌點數?如何去預估可以達到更好的效果?這些問題常常會從以前的經驗來去評估隔天或者未來一段時間內股票的指數，一旦評估錯誤可能會導致一定程度的虧損或決策上的失誤，所以股票個股的評估成為每日在收盤後重要的議題。
本研究將以倒傳遞類神經網路(Back Propagation Network)並結合統計的方法來進行分析和預估未來股票及個股的收盤指數，並用來預估未來一定時間內之股市收盤、個股收盤，使用均方根誤差率(RMSE)及日平均誤差來比較分析結果好壞，本研究利用台股指數1999年到2010年的收盤指數以及2005年到2010年的台積電、華碩兩檔個股來進行測試，並針對倒傳遞類神經訓練資料及預估時輸入資料做了實驗並由實驗結果得知運用倒傳遞類神經網路結合統計的方法來預估可以達到一定的效果。

關鍵字

加權移動平均法 ； 股票預測 ； 倒傳遞類神經

並列摘要

The direction of index in the stock market is a topic often get concerned by people. Often hear about people's discussion on how to evaluate the trend of next day index, and how to assess to achieve a better result? people usually assess the stock market trend by previous trend, once an improper assess occurred can lead to certain degree of financial loss, therefore, assessment on individual stocks become an important issue after the stock market close.
In this research, we use Back Propagation Network method combined statistics to measure up the analysis on the close value of stock market, and ndividual stocks and predict the trend of stock and individual stocks index in a period of time, also use RMSE and daily average error to analyze result. During this research, we use the close value of Taiwan Stocks Index from 1999 to 2010, and for the individual stock, we choose Taiwan Semiconductor Manfacturing Company (TSMC) and ASUS to perform the test. Based on the training data of the Back Propagation Network and prediction data during this experiment result, we find out use Back Propagation Network method combined statistics can achieve certain degree result on assessment.

並列關鍵字

Back Propagation Network ； Stock Prediction ； weighted moving average

參考文獻

[樓克望 2011]樓克望，GARCH模型與條件分配對俄羅斯股市波動性預測能力之研究，國立臺北大學國際財務金融碩士論文。

[王奐霖2012] 王奐霖，應用倒傳遞類神經網路和多因子軟體專案分群預估軟體工作量，2012。

[Chen 2011] Shyi-Ming Chen and Chao-Dian Chen, “TAIEX Forecasting Based on Fuzzy Time Series and Fuzzy Variation Groups”， IEEE Transactions on Fuzzy Systems, Vol. 19, No. 1, February 2011.

[Chen 2010] Shyi-Ming Chen and Nai-Yi Wang, “Fuzzy Forecasting Based on Fuzzy-Trend Logical Relationship Groups”, IEEE Transactions on Systems, Man, and Cybernetics—Part B: Cybernetics, Vol. 40, No. 5, October 2010.

[Wong 2010] Wai-Keung Wong, Enjian Bai, and Alice Wai-Ching Chu, “Adaptive Time-Variant Models for Fuzzy-Time-Series Forecasting”, IEEE Transactions on Systems, Man, and Cybernetics—Part B: Cybernetics, Vol. 40, No. 6, December 2010.