Candlestick technical analysis is an old trading technique that tracks the short-term price movements by employing the relationship between open, high, low, and close prices. The purpose of this thesis is to examine the predictive power of candlestick trading strategies by using the Taiwan 151 component stocks daily data for the period from 2 January 1992 to 31 December 2009. The main contribution of this thesis is using a fourprice- level approach to categorize the single-line patterns constructed by candlestick charting in a systematical manner. The approach adopted in this thesis permits us to release for the limitation of recognition in a manner not previously possible. Moreover, we not only consider transaction costs and risk but also mitigate data-snooping problems conscientiously by several appropriate methods, including the bootstrap methodology and sub-sample and out-of-sample tests. We find evidence that four patterns are profitable for the Taiwan stock market after transaction costs, including one bullish pattern and three bearish ones.

K線技術分析是一種歷史悠久的交易技巧，它利用開盤價、最高價、最低價與收盤價來追蹤短期價格的變動。本文的目的是以台灣加權指數151支成份股，1992年1月2日到2009年12月31日的日資料為樣本來檢驗K線交易策略的預測能力。最主要的貢獻是本文使用一個系統化的檢驗方式-四價分析法，這方法可以將單根K線完整地分類。這個方法也解除了過去對K線型態認定上的限制。另外，我們不只考慮了交易成本與風險，也用了數種適當的方法來解決資料探勘的問題，包括拔靴法以及切割樣本和樣本外檢驗。本文的結論發現，在扣除交易成本後，有四種單根K線確實在台灣股市具有獲利能力，其中包含一個買進訊號與三個賣出訊號。