股價操縱旨在使價格偏離基本價值並伺機從中獲利事影響資本市場運行、投資人權益與公共利益甚鉅。現存文獻中股價操縱的理論模型眾多事然而相闋的預警卻一直未有適當的發展。本文嘗試根據Phillips and Yu（2011）的右尾遞迴式草根檢定方式提出一個新的股價操縱預警方式今其利用不同的操縱手段中使股價呈現緩步上漲（running-up）的共同現象做為認定條件事以彌補學術文獻的不足並精進監理實務中的相關做法。本文的警示方法不僅具有前瞻性、即時性等優點事同時僅需要股價資料而使偵測成本低，並可以明確地定義出操縱期間的起龍點。我們以1994年至2010年之間違反證券交易法第155條業經法院裁定股票操縱的案例公司為樣本事研究證實本文的操縱預警效果可補足現行的注意股票公告；同時事其所認定的股價操縱期間與法院審理所判定的期間也有相似或可供比較的表現。我們進一步根據所認定之操縱期鬧事對所有案例分別在操縱佈局、價格上漲、價格崩跌、後操縱期等四個期間中，解構出股價操縱橫斷面各個階段的股價的特徵與市場品質事並驗證不同期間的差異性。重要的實證發現如下：（1）評估風險中常用的GARCH模型並無法適切地反應股價業已遭受操縱的風險。因此事本文可以前瞻性地預警股價操縱的方法益發顯得重要。（2）操縱期間的上漲階段與崩跌階段之期間長短呈現嚴重不對稱的情況；而股價操縱上漲期間最高點價格平均而言為起漲前價格的三倍。（3）利用操縱期間報酬率可以概估出操縱者的潛在最大可能獲利、以及被操縱者最大可能損失等事並從中得出不同的獲利或停損的進、退場時機的意涵。本文的方法可以便利地為公開資訊觀測站所採周編制，定期發布做為偵測股價操縱的預警模式今不但可以防範有心人的操弄、提升交易訊息透明度與降低資訊不對稱，保護資訊相對弱勢的股民今還可以讓法院與檢調體系對金融、經濟犯罪的審理多了一項可供使用的科學工具。長期而言，對本國的法治以及促進資本市場健全發展有相當重要的制度設計性意義。

Stock price manipulation will erode market efficiency, ruin trading orders in capital markets, and bring undesired social cost. How to detect it early has long been an interesting and important issue for academics and supervisory authorities. While we have many theoretical models deducing insightful policy implications over the past decades, empirical research is limited due to data availability. In this paper, we propose an early warning model to detect equity price manipulation using the recursive right-tailed unit root test suggested by Phillips and Yu (2011). As we aim to detect the existence of price manipulation instead of testing any specific type of manipulation, we suggest identifying manipulation via the common phenomenon of running up prices that prevails among the vast amount manipulation strategies. By retrieving the sentenced cases of price manipulation from the Law and Regulation Retrieving System of the Judicial Yuan of R.O.C. within the period from 1994 to 2010, we examine the price data and identify the manipulation period using the new technique. The ability of this model to identify manipulation is admirable when contrasted to the manipulation periods identified by the court judges. Our approach is also comparable to the scheme ＂Attention Securities Announcement＂ published by the TWSE and OTC in featuring the risk of price manipulation. We further categorize each manipulation period into 4 stages and disentangle a set of comprehensive stylized facts in understanding the patterns of pump and dump manipulation in Taiwan. Several findings are worth noticing: (1) The commonly employed GARCH model for risk evaluation does not reveal whether the equity has been manipulated. (2) The average pumping period is about 2.4 times longer than the average dumping period, and the highest pumped price can be on average 3 times as high as the initial price before pumping. (3) We can characterize the maximal possible manipulation profits and maximal possible loss for investors by the cross-sectional average holding-period returns for the pumping and dumping periods, respectively. Our easy-to-implement manipulation monitoring approach is shown to be forward looking and effective. Proper installation and implementation of it for public information disclosure may help enhance the transparency and quality of the market, deter manipulation activities, and protect uninformed investors.