本論文分析臺灣期貨市場當沖交易的交易概況，探討不同類別投資人的獲利能力，以及影響當沖報酬的因素。研究結果顯示，台灣期貨市場的當沖交易非常活絡，平均當沖交易占總交易量的74.71%，而個別投資人為主要的參與者，占了當沖交易的66.68%。而先賣當沖可獲得比先買當沖更多的報酬，40.16%的先賣當沖，其當沖報酬可以超越當日的市場報酬。在交易獲利方面，整體而言，平均每筆當沖交易投資人會有-111.55的虧損，以投資人類別分類後可以發現，和股票市場有很大的不同，有52.23%的個別投資人可以在當沖交易中獲利，而投信只有21.04%。整體投資人績效方面，在考慮交易成本後，只有外資可以獲得正報酬，相較其他投資人具有資訊優勢，而自營商和投信的績效則比個別投資人差，並不具有資訊優勢。 交易量分析方面，平均而言，在研究期間內交易量小於10筆的投資人可以獲得較高的報酬，而交易量大於10000的投資人則有小幅度的虧損。在單獨分析波動度較大的交易日中，發現投資人在這些交易日中可以獲得比平常更多的報酬。我們進一步去分析在研究期間內績效表現較好的投資人，發現他們平均的交易量只有12.65筆，這類投資人並不傾向先買和先賣當沖，但是傾向一天只從事一種當沖交易，但先賣當沖的報酬仍然會多於先買當沖。這些投資人的報酬有集中於波動度大的交易日的趨勢，而交易時間比般人的交易時間要來的長，平均一筆交易為91到96分鐘，而整體投資人的交易時間約為48到50分鐘。 在迴歸模型中，我們發現市場報酬、交易量、時間、投資人過去績效對於當沖報酬都有顯著的解釋能力，然而，交易時間對於報酬卻是負向的關係。在解釋力方面，先賣當沖的解釋力較先買當沖好，而超額報酬模型的解釋力較當沖報酬模型好。 / When an investor buys and sells the future on the same day, he was made a day trade. This thesis studies the performance of day trades in the Taiwan Furtures Exchange(TAIEX) for the period between July 2006 and December 2006. Day trading accounts for 74.71% of total trading activities, of which individual investors account for 66.68% of the day trading volume. The results showed that sell and then buy day trading can earn more profits than buy then sell day trading, about 40.16% of the sell and then buy day trading can earn more profits than market return. In aggregate, day traders earn mean gross profits(before transaction costs) of $NT 199 per transaction, but net losses(after a reasonable accounting for transaction costs) of $NT 111 per transaction. After we categorize day traders, we find that about 52.23% of the indiviual investers can profit by day trading activities, and only foreign investors can earn sufficient profits to cover transaction costs. To analyze the performed better investers, we find that their average trading volume only have 12.65 within half year, they don’t tend to sell and then buy day trading or buy then sell day trading, but they tend to engage one kind of trading in one day, and sell and then buy day trading can earn more profits than buy then sell day trading. Their trading time is longer the others, about 91 to 96 minutes, other traders’ trading time only about 48-50 minute. Overall, the evidence suggests that market returns、trading volume、trading time and past profitablilty have a strong relation with subsequent returns, day traders who historically earned profits continue to earn profits. The statistical power of sell and then buy regression model is stronger than buy then sell regression model.

¡@¡@With high liquidity and operating flexibility, futures can not be held for long term and must be restricted by margin requirement. This makes many futures traders prefer day trading to avoid the risk of the price gap of the next trading day. ¡@¡@Day traders tend to operate a trend-following strategy based on technical analysis and actively manage their holding positions. They take stop-loss strategy in the wrong direction to limit the damage, while take raise-stake strategy in the right direction to increase profits. A program trading system can even be utilized to carry out the strategy immediately and mechanically. ¡@¡@This study use Taiwan Stock Exchange Capitalization Weighted Stock Index Futures as our sample. It simulates the performance of trend-following strategy of day traders by using moving average as basic signals, combined with the trading volume and Bollinger Band. ¡@¡@We found that, in medium-term and long-term time frame (30-60 minutes), active management of stop-loss and overweight strategy can still change the distribution of performance and earn positive returns, even if moving average technical analysis is not as effective as expected.

Thesis (M.S.)--Duquesne University, 2006. / Title from document title page. Abstract included in electronic submission form. Includes bibliographical references (p.28-29) and index.

The dissertation consists of two essays. In the first essay, we measure herding by institutional investors in the new economy (internet) stocks during 1998-2001 by examining the changes in the quarterly institutional holdings of internet stocks relative to an average stock. More than 95% of the stocks that are examined are listed on NASDAQ. The second essay attempts to detect intra-day herding using two new measures in an average NYSE stock during 1998-2001. In the second essay, rather than asking whether institutional investors herd in a specific segment of the market, we endeavor to ask if herding occurs in an average stock across all categories of investors. The first essay analyzes herding in one of the largest bull runs in the history of U.S. equity markets. Instead of providing a corrective stabilizing force, banks, insurance firms, investment companies, investment advisors, university endowments, hedge funds, and internally managed pension funds participated in herds in the rise and to a lesser extent in the fall of new economy stocks. In contrast to previous research, we find strong evidence of herding by all categories of institutional investors across stocks of all sizes of companies, including the stocks of large companies, which are their preferred holdings. We present evidence that institutional investors herded into all performance categories of new economy stocks, and thus the documented herding cannot be explained by simple momentum-based trading. Institutional investors' buying exerted upward price pressure, and the reversal of excess returns in the subsequent quarter provides evidence that the herding was destabilizing and not based on information. The second essay attempts to detect herding in financial markets using a set of two methodologies based on runs test and dependence between interarrival trade times. Our first and the most important finding is that markets function efficiently and show no evidence of any meaningful herding in general. Second, herding seems to be confined to very small subset of small stocks. Third, dispersion of opinion among investors does not have much of impact on herding. Fourth, analysts' recommendations do not contribute to herding. Last, the limited amount of herding on price increase days seems to be destabilizing but on the price decrease days, the herding helps impound fundamental information into security prices thus making markets more efficient. Our results are consistent with Avery and Zemsky (1998) prediction that flexible financial asset prices prevent herding from arising. The seemingly contradictory results of the two essays can be reconciled based on the different sample of stocks, and the different methodologies of the two essays which are designed to detect different types of herding. In the first essay, herding is measured for NASDAQ-listed (primarily) internet stocks relative to an average stock, while the second essay documents herding for an average stock. In the first essay, we document herding in more volatile internet stocks, but we do not find any evidence of herding in more established NYSE stocks. The first essay examines herding by institutional investors, while the second essay examines herding, irrespective of the investor type. Consequently, in the first essay, we find that a subset of investors herd but in the second essay market as a whole does not exhibit any herding. Moreover, the first essay measures herding by examining the quarterly institutional holdings of internet stocks, while the second essay measures herding by examining the intra-day trading patterns for stocks. This suggests that it takes a while for investors to find out what others are doing leading to herding at quarterly interval but no herding is observed at intra-day level. The evidence presented in the two essays suggests that while institutional investors herded in the internet stocks during 1998-2001, there was very little herding by all investors in an average stock during this period. / Ph. D.

Abstract This thesis investigates a set of equity market phenomena associated with investors' trading activity, using a comprehensive Finnish Central Securities Depository (FCSD) database that records practically all trades by Finnish investors. This database enables us to classify a large number of heterogeneous investors using both economic and institutional characteristics. The first essay classifies investors by trading activity. It analyzes trading styles of active and passive investors during the boom in technology stocks 1997–2000. We find that the herding tendency of active investors grew monotonically, year by year. Particularly large active investors used momentum and growth strategies. Moreover, buy pressures of active investors were positively related to contemporaneous daily returns. Passive investors, on the other hand, herd very strongly and their trading exhibited a contrarian style throughout the sample period. The second essay focuses on the relation between day trading of individual investors and intraday stock price volatility. I find a strong positive relation between the individual investors' day trades and volatility for actively day traded stocks. This finding suggests that day trading tends to increase volatility and/or day traders tend to become more active on the days of high volatility. The third essay tests the theoretical proposition of Amihud and Mendelson (1986) that investors hold assets with higher bid-ask spreads for longer periods. We measure holding periods of individual investors directly and find that they are positively related to spreads. The models control for a variety of other stock characteristics (e.g. value vs. growth orientation) and investors' attributes (e.g. gender) affecting holding periods. The fourth essay studies how both individual and institutional investors with different levels of capital gains and losses react to earnings announcements. I find that both sign and magnitude of capital gains affect individual investors' abnormal trading volumes. Individual investors are less prone to sell when they are carrying loses rather than gains. Furthermore, they react less to earnings announcements when capital gains or losses are large (over 20%). Taken together these findings provide support for prospect theory. Institutional investors appear to be less affected by psychological factors underlying prospect theory.

Stock trading provides a relatively low risk and hassle free channel for people to invest, when compared with other methods. However, it is unfortunate that, for those other people who may be equally interested and financially able in stock trading, they have yet to experience the simplicity of, and the wealth accumulated from stock investments. Reasons for their hesitation to be involved in stock trading vary; but primarily it is due to their lack of experience in trading.

<p>IPO underpricing is a phenomenon found in all markets worldwide. Investors are always looking for a good opportunity of short-term abnormal positive returns, and the IPOs first-day trading returns have been a good investment strategy for both institutional and private investors in all markets of the world.</p><p>This study consists at an investor’s perspective analysis of the first-day returns of 59 IPOs listed on the Brazilian Stock Exchange Market from January 2004 to April 2007, where I have found a significantly mean positive underpricing of 6,60%.</p><p>I have found also some evidences of a sprouting “hot-market” period in Brazil, since the number of the IPOs in Brazil has been growing almost in an exponential speed, taking advantage of the constant growing cash inflow and liquidity of the Brazilian market, followed by the high evaluation of the Ibovespa Index, with return of 140% on the study time frame.</p><p>When categorizing the study by year, by underwriter (investment bank) and by market segment, I always have found positive adjusted initial returns, which corroborates the fact that underpricing is a constant phenomenon in the Brazilian market.</p><p>Other important facts that were identified in this study is that the average returns of the IPOs are decreasing along the years and that companies that depend to a large extent on their human capital and are in the business areas that are staff intensive have a high level of underpricing while companies that have a high level of fixed assets have a low level of underpricing.</p><p>Finally, after performing a multivariate linear regression analysis with the chosen independent variables on the full sample and some categorized samples, the results did not have enough statistical significance and consistence that could make them useful to create a statistical model to explain the underpricing level of Brazilian IPOs between January 2004 and April 2007.</p>

IPO underpricing is a phenomenon found in all markets worldwide. Investors are always looking for a good opportunity of short-term abnormal positive returns, and the IPOs first-day trading returns have been a good investment strategy for both institutional and private investors in all markets of the world. This study consists at an investor’s perspective analysis of the first-day returns of 59 IPOs listed on the Brazilian Stock Exchange Market from January 2004 to April 2007, where I have found a significantly mean positive underpricing of 6,60%. I have found also some evidences of a sprouting “hot-market” period in Brazil, since the number of the IPOs in Brazil has been growing almost in an exponential speed, taking advantage of the constant growing cash inflow and liquidity of the Brazilian market, followed by the high evaluation of the Ibovespa Index, with return of 140% on the study time frame. When categorizing the study by year, by underwriter (investment bank) and by market segment, I always have found positive adjusted initial returns, which corroborates the fact that underpricing is a constant phenomenon in the Brazilian market. Other important facts that were identified in this study is that the average returns of the IPOs are decreasing along the years and that companies that depend to a large extent on their human capital and are in the business areas that are staff intensive have a high level of underpricing while companies that have a high level of fixed assets have a low level of underpricing. Finally, after performing a multivariate linear regression analysis with the chosen independent variables on the full sample and some categorized samples, the results did not have enough statistical significance and consistence that could make them useful to create a statistical model to explain the underpricing level of Brazilian IPOs between January 2004 and April 2007.

本文採用1989年10月2日至2017年4月12日的台灣股市加權指數日資料，並分為漲跌幅限制為7%的區間一，以及放寬為10%的區間二。接著使用〔(最高價-最低價)/昨日收盤價〕以及〔(收盤價-開盤價)/昨日收盤價〕兩個不同變數來衡量台灣股市單日的波動與報酬，然後也運用了週轉率、成交金額、5日均值比三種方法來估算股市成交量。藉此探討台灣股市波動與成交量的關係。使用的方法是分量迴歸模型，更細部的研究股市上漲或下跌時，每個分量之下不同的價量關係。 實證結果顯示，台灣股市普遍存在「價漲量增」與「價跌量增」的現象，且在波動越大的時候也就是分量尾端的部分，其關係更加的明顯。另外，使用三種變數來衡量成交量，在區間二大致得出相同的結論，但是區間一因為週轉率與成交金額的歷史走勢具有差異，所以結果也不盡相同。但是使用週轉率在歷史樣本中更具有相同的比較基礎，因此得出的結論也較一致，所以認為台灣股市仍是以「價漲量增」與「價跌量增」為普遍現象。 / This paper used the Taiwan stock market index daily data from October 2, 1989 to April 12, 2017, which divided into a range of 7% of the price limit, and a range of 10%. There are two different variables to measure the volatility and return: [(the highest price - the lowest price) / yesterday's closing price] and [(closing price - opening price) / yesterday's closing price], and three different variables: turnover, dealing amount, 5-day average ratio to estimate the stock market volume. The method used is quantile regression model, and that allows us to observe different relationship between volatility and volume under every single quantile. Empirical results show that there are two phenomena exist in the stock market of Taiwan: "rising values increase in volume" and "falling values increase in volume." In addition, the use of three variables to measure the volume, in the interval 2 roughly come to the same conclusion, but in the interval 1 because the historical trend of turnover rate and dealing amount are different, so the results are not the same. But the use of turnover in the history sample has the same comparison basis, so the conclusions are more consistent, so that the Taiwan stock market is still the " rising values increase in volume " falling values increase in volume" as a common phenomenon.

Submitted by Marcio Pintan (mpintan@gmail.com) on 2018-06-25T18:13:19Z No. of bitstreams: 1 financas-pintan-dissertação versao final.pdf: 1855884 bytes, checksum: 0ac7f69afc81d4cea8446f01d309fd99 (MD5) / Approved for entry into archive by Mayara Costa de Sousa (mayara.sousa@fgv.br) on 2018-08-13T16:18:19Z (GMT) No. of bitstreams: 1 financas-pintan-dissertação versao final.pdf: 1855884 bytes, checksum: 0ac7f69afc81d4cea8446f01d309fd99 (MD5) / Approved for entry into archive by Suzane Guimarães (suzane.guimaraes@fgv.br) on 2018-08-13T16:25:32Z (GMT) No. of bitstreams: 1 financas-pintan-dissertação versao final.pdf: 1855884 bytes, checksum: 0ac7f69afc81d4cea8446f01d309fd99 (MD5) / Made available in DSpace on 2018-08-13T16:25:32Z (GMT). No. of bitstreams: 1 financas-pintan-dissertação versao final.pdf: 1855884 bytes, checksum: 0ac7f69afc81d4cea8446f01d309fd99 (MD5) Previous issue date: 2018-05-25 / Esta dissertação trata das operações realizadas na BM&F BOVESPA chamadas comumente de 'Day Trading', ou seja, operações cuja compra (ou venda) e a liquidação são realizadas no mesmo dia. Tal questão é relevante, principalmente para o pequeno investidor, por possibilitar a otimização do resultado da sua carteira de investimento ao longo do tempo. O objetivo de pesquisa deste trabalho é apresentar e testar algumas técnicas utilizadas pelos operadores do mercado financeiro na modalidade 'Day Trading'. Em conjunto com a verificação das teorias de análise gráfica, o trabalho pretende conciliar tais técnicas preditivas com teorias de gestão de risco e de gestão de portfólio, nesse caso mais precisamente a teoria moderna de portfólio de Markowitz, de forma a testar a eficiência da combinação entre essas teorias no mercado de ações brasileiro, e se existe a possibilidade de otimização dos resultados que um investidor pode alcançar ao longo do tempo. Para atingir este objetivo foi realizada uma pesquisa quantitativa utilizando técnicas de análise gráfica baseadas em teorias amplamente conhecidas no mercado de capitais, como os Princípios de Ondas de Elliott e a Teoria de Dow. A partir dos indicadores de sucesso obtidos por essas técnicas preditivas (através de 'backtests'), o presente trabalho testa a efetividade das questões relativas a eficiência de mercado apresentadas nas Hipótese de Mercados Eficientes de Fama (1970). As principais conclusões desta dissertação sugerem que uma estratégia passiva, de compra e manutenção do Índice Bovespa, domina respectivamente estratégias baseadas na Teoria de Markowitz e estratégias ativas de Day Trading baseadas em análise técnica. Os resultados trazem uma grande contribuição para o pequeno investidor através de uma maior compreensão sobre possibilidades que as operações de curto prazo podem trazer para ao seu portfólio de investimentos e confirma a visão de que o mercado de ações brasileiro é eficiente em sua forma fraca. / This thesis deals with operations carried out on BM&F BOVESPA commonly called 'Day Trading', which are operations whose purchase (or sale) and settlement are carried out on the same day. This issue is relevant, especially for the small investor, because it allows the optimization of the result of their investment portfolio over time. The objective of this research is to present and test some techniques used by financial market traders in the 'Day Trading' modality. In conjunction with the verification of theories of technical analysis, the paper aims to reconcile such predictive techniques with theories of risk management and portfolio management, in this case more precisely the Modern Portfolio Theory of Markowitz, in order to test the efficiency of the combination between these theories in the Brazilian stock market, and whether there is a possibility of optimizing the results that an investor can achieve over time. To achieve this goal a quantitative research is performed using graphical analysis techniques based on theories widely known in the capital market, such as the Elliott Wave Principle and the Dow Theory. From the indicators of success obtained by these predictive techniques (through backtests), the present research explores aspects of market efficiency presented in the Efficient Market Hypothesis of Fama (1970). The main conclusions of this dissertation suggest that a passive buy and hold strategy of the Bovespa Index dominates respectively strategies based on Markowitz Theory and active day trading strategies based on technical analysis. The results make a contribution to the small investor through a better understanding of the possibilities that short-term operations can bring to their investment portfolios and confirms the view that the Brazilian stock market is efficient in its weak form.