標題: 運用類神經網路於臺灣證券市場基本面分析
Using Neural Networks On A Fundamental Analysis In The Taiwan Stock Market
作者: 呂玉銘
Leu, Yue-Ming
陳安斌
Chen, An-Ping
資訊管理研究所
關鍵字: 基本面分析;證券市場;內線交易;類神經網路;財務報表;Fundamental Analysis;Stock Market;Inside Trading;Neural Networks;Financial Statement
公開日期: 1993
摘要: 隨著經濟的成長,臺灣證券市場上一股投資熱潮;然而,投資大眾往往只偏重於股市技術面的分析,而忽略了能真正反應公司本質的基本面。殊不知道投資的基本精神實在於技術面、基本面,更甚而總經面,政治情勢與國際政局變化等全方位的考量。但不管如何,影響上市公司股價漲跌最明顯的,要算是公司的財務狀況了。因此,本研究嘗試利用類神經網路,來掌握公司財務變化股價漲跌幅度的影響。 故最初本研究假設上市公司所公佈的財務季報表狀況,會反映在財務報表公佈後的股價漲跌上面。而根據研究的結果顯示:股價漲跌趨勢,隨著上市公司財務報表公佈後來預測其準確度平均為54%。故本研究嘗試假設:股價漲跌趨勢,將會反映在財務報表公佈之前,而研究結果顯示其預測準確度平均達64%,甚至有公司其準確度高達80% 。因此由本研究的結論,我們可以確信:臺灣證券市場仍是一個不公平的市場,因為其中實隱含著可能的內線交易與不正常的投機行為。
Along with the booming of economy in Taiwan, there are so many people put their money into the stock market. However, there are greater part of investors always keep their eyes on the technical analysis and lose sight of the fundamental analysis which show the actual essence about a company. They don’t know that the basic spirit of investment is to aim at technical side, fundamental side, even the macro economy and political sides. However, the thing that affects the range of stock price obviously, is the financial condition of a company. Therefore, this study is trying to use neural networks to master the effect of any major financial changes to stock price. This study assumes that stock price rising or falling depends on the financial statements as they are published. Following this study, we will learn that we can predict the stock price right after we finish publishing the financial statements. But the accuracy is about 54%. We try to assume that stock price will be reflected is about 54%. We try to assume that stock price will be reflected as the financial statements are published, and the accuracy is about 64% in average; the accuracy is sometime up to 80%. To conclude, the stock market in Taiwan is a pretty unstable one that implies an illegal inside trading and a speculative enterprise.
URI: http://140.113.39.130/cdrfb3/record/nctu/#NT823396025
http://hdl.handle.net/11536/58626
顯示於類別:畢業論文