本研究的目的是嘗試建立一個結合個別技術指標與類神經網路的應用系
統,而這系統的目的在於:1. 能夠解決技術指標鈍化與相互衝突之現象.2.
能夠解決技術指標在不同期間不一致表現之現象.3. 藉由上述二問題之解
決,使本系統能夠提供買賣決策供投資人參考,讓投資人能提高 其獲利.
經過四個外匯匯率以兩種不同輸入組合之實證後,整體而言,本系統可達到
穩定獲利之目標,對於技術指標在不同期間表現不一致現象之解決成效良
好,但對於指標間衝突與鈍化現象只能稍做改善.就投資報酬率而言,本系
統之投資報酬率平均高達40%以上,因此,本架構可成為投資人制定投資決
策的輔助工具,藉由此架構,可以提高投資人之投資績效.

In this thesis, we investigate the investment performance in the
foreign exchange, through empirical data analysis of neural
network andtechnical indices. The main purposes of this module
include:1. To solve the following problems of technical analysis
indices signal insentsitive and conflict.2. To solve the problem
of technical analysis indices which have inconsistent
performance in different periods.3. By these two functions, this
module will provide valuable reference to investors, so that
they will get higher revenue.Our results show that neural
network and technical analysis can make revenue more stable and
can successfully solve the problem of technical analysis
indiceswhich have inconsistent performance in different periods.
We show that, this moudle can improve technical indices signal
insensitive and conflict.