Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Overnight Stock Price Reversals

Overnight Stock Price Reversals In present study, I explore the dynamics of stock price reversals. In particular, I try to shed light on the overnight reversals, that is, on the price reversals between the end of a trading day and the opening session of the next trading day. To account for the "end-of-the-day" price moves, for each of the stocks currently making up the Dow Jones Industrial Index, I compare, on the daily basis, the high-to-close and the low-to-close price changes, and also compare them to the same day's average and median changes for the total sample of stocks. I document that opening returns tend to be higher following the days with relatively large high-to-close price changes (price decreases at the end of the day), and lower following the days with relatively large low-to-close price changes (price increases at the end of the day). Such "overnight reversals" price behavior seems to contradict the market efficiency. Finally, I construct five portfolios based on the opening trading sessions and involving a long position in the stocks on the days when, according to the "overnight reversals" behavior, their opening returns are expected to be high and a short position in the stocks on the days when their opening returns are expected to be low. All the portfolios are found to yield significantly positive returns, providing an evidence for the practical applicability of the "overnight reversals" pattern in stock prices. Keywords: Intraday Stock Prices; Opening Stock Returns; Overreaction; Stock Price Reversals JEL Classifications: G11, G14, G19. 1. Introduction In informationally efficient asset markets, prices incorporate news quickly and accurately and investors cannot predict future returns and make abnormal profits. However, the empirical results of Shiller (1984), and DeBondt, and Thaler (1985) seriously challenge the notion of efficient capital markets and indicate that abnormal profits are possible using historical information. More specifically, they demonstrate that a `contrarian' strategy of going long a portfolio of extreme prior losers and going short a portfolio of extreme prior winners will produce long-term abnormal profits. This finding is explained the tendency of investors to overreact to information, resulting in subsequent price reversals. Since these pioneering studies, a large volume of theoretical and empirical research work has analyzed price overreaction in financial markets. A number of potential explanations for this phenomenon have been suggested, the most popular of them including bid-ask biases, investor psychology, multifactor pricing models, size and transaction costs. For example, short-term reversals may be induced by prices bouncing between bidask quotes. Jegadeesh, and Titman (1995) show that inventory imbalances may cause negative short - term serial correlation in prices, while Roll (1984) shows that due to the dealers order processing costs the bid-ask bounce may explain short - term negative serial correlation (for bid - ask explanations, see also, Cox, and Peterson 1994, Atkins, and Dyl 1990; Park 1995). Fama, and French (1996) find that long-term equity return reversals can be explained within the context of a multifactor asset pricing model, while Zarowin (1990) finds that contrarian profits may be due to a size effect in stock returns. Other authors attempt to explain overreaction by means of investor psychology (e.g., Barberis et al. 1998; Daniel et al. 1998; Lakonishok et al. 1994; Hong, and Stein 1999; Odean 1998), and present different channels through which investor psychology can lead to inefficiencies in securities' returns. The rationale for these studies originates from evidence of empirical psychology that individuals tend at times to underreact or overreact (Kahneman, and Tversky 1982; Griffin, and Tversky 1992). For example, the confidence model provided by Daniel et al. (1998) predicts long-run negative serial correlation, and the model developed by Barberis et al. (1998) argues that investors overreact to strong and salient information. The latter model is based on two well - Journal of Advanced Studies in Finance established human psychological characteristics: representativeness and conservatism, and suggests that overreaction is expected to cause future reversals as prices revert to their fundamental value. In order to distinguish stock price overreaction and market inefficiency from predictable changes in expected returns, Lehmann (1990) suggests examining returns over short time intervals. In fact, the focus on long-term dynamics in stock returns in the papers by Shiller (1984), and De Bondt, and Thaler (1985) is more recently realigned to short - run return behavior, ranging over time periods from a few days up to a month, in the major part of the subsequent literature (e.g., Zarowin 1989; Atkins, and Dyl 1990; Cox, and Peterson 1994; Park 1995; Bowman, and Iverson 1998; Nam et al. 2001). The major focus of these studies is on identifying potentially profitable contrarian strategies built on a reverting behavior of stock prices in the short run. For example, Lehmann (1990), and Jegadeesh (1990) show that contrarian strategies that exploit the short-run return reversals in individual stocks generate abnormal returns of about 1.7% per week and 2.5% per month, respectively. Remarkably, Conrad et al. (1994) document that reversal profitability increases with trading activity. A continuously growing body of literature concentrates on even shorter time intervals, and studies intraday price reversals. Zawadowski et al. (2006) analyze a large sample of NASDAQ-traded stocks, and define "large intraday price changes" as intraday price changes bigger than a certain level of 2-6% price change within 10-120 minutes, and alternatively, as intraday price changes exceeding 6-10 times the normal volatility during that time of the day. They document significant stock price reversals following these initial changes, implying that the latter contain at least some element of overreaction. Grant et al. (2005) look at large changes in the prices of S&P 500 futures during the opening trading session (with respect to the previous day's closing price), and find that such changes are followed by significant price reversals in the first half-an-hour to hour after the market opening. The main goal of the present study is to shed a little more light on the dynamics of stock price reversals and namely, on the overnight reversals. The study is logically connected to the study by Grant et al. (2005), but looks at the "opposite" order of events and analyzes a more general case. Namely, I expect that following some relatively large price moves towards the end of the trading days, there may be price reversals during the next days' opening sessions1. To account for price changes at the end of the trading days, I employ high-to-close and low-to-close price differences. The use of these measures is in line with Becker et al. (2008), and Klossner et al. (2012), who employ them as integral parts of their intraday upside and downside volatility measures, respectively. The focus on the next days' opening sessions as potential "reversal periods" is motivated by the short-time nature of the reversals, in general, and by the observation (e.g., Harris 1986) that the predominant portion of stock price moves takes place within the first minutes of trading. I analyze intraday price data on thirty stocks currently making up the Dow Jones Industrial Index, and find supporting evidence for my research hypothesis. For each trading day, I compare each stock's high-to-close and low-to-close price changes, and also compare them to the same day's average and median changes for the total sample of stocks, and document that opening returns tend to be higher following the days with relatively large high-to-close price changes (price decreases at the end of the day), and lower following the days with relatively large low-to-close price changes (price increases at the end of the day). These findings imply that stock price changes towards the end of a trading day may contain an element of overreaction to be reversed right at the beginning of the next trading day. Based on these findings, for the opening trading sessions, I construct a number of daily-adjusted portfolios involving a long position in the opening session in the stocks on the days when, according to the findings, their opening returns are expected to be high and a short position in the opening session in the stocks on the days when, according to the findings, their opening returns are expected to be low, and demonstrate that the (opening) returns on these portfolios are significantly positive. The rest of the paper is structured as follows: In Section 2, I describe the data sample. Section 3 comprises the research hypotheses and the results. Section 4 concludes. 2. Data description For the purposes of present research, I employ daily opening, high, low and closing prices of thirty stocks currently making up the Dow Jones Industrial Index over the period comprised from January 2, 2002 to September 30, 2011 (overall, 2456 trading days).2 I adjust all the prices to dividend payments and stock splits, and for each stock in the sample and for each trading day in the sampling period, calculate: I test this hypothesis against the opposite one, suggesting that in the opening sessions, stock prices may continue to move in the direction of the respective changes that took place towards the end of the previous days. 2 The data were taken from the Yahoo Finance website. 1. Stock's opening return (i.e., stock price's change from last day's closing price to today's opening price) as: RO,it PO,it PC ,it (1) where: RO,it is stock i's opening return on day t; PO ,it is stock i's opening price on day t; and PC ,it is stock i's closing price on day t-1. 2.Stock's daily return (i.e., stock price's change from last day's closing price to today's closing price), as: RD ,it PC ,it PC ,it (2) where: R D ,it is stock i's daily return on day t. 3.Stock's high-to-close price difference, as: R HC ,it PH ,it PC ,it (3) where: RHC,it is stock i's high-to-close price difference on day t; and PH ,it is stock i's highest price on day t. 4.Stock's low-to-close price difference, as: RLC ,it PC ,it PL ,it (4) where: is stock i's low-to-close price difference on day t; and PL,it is stock i's lowest price on day t. Of course, since the following relations between the intraday prices hold: PH ,it PC ,it ; and PL,it PC ,it (5) the last two price differences are defined so that they are non-negative, representing the absolute values of the respective price changes. Table 1 comprises the basic descriptive statistics of the intraday price differences and returns for the thirty sample stocks. At this stage, we may note that, as it might be expected for the largest industrial companies of the US, 27 out of 30 stocks have positive mean daily returns, the remaining 3 showing negative, yet close to zero daily returns. Overall, the mean daily returns range from -0.005 to 0.076 percentage points, with standard deviations ranging from 1.183 to 3.568 percentage points. The two intraday mean price differences are highly correlated in the cross-section, that is, all of them are relatively high for certain stocks and relatively low for other ones. One more thing to note is that for 23 out of 30 stocks, the mean low-to-close differences are greater than the mean high-to-close differences. 3. Research hypotheses and results 3.1. Effect of the "end-of-the-day" stock price moves on the next day's opening returns The concept of stock price reversals is well-documented and widely-discussed in financial literature. Many studies show that stock prices often overreact to news and subsequently revert themselves in order to arrive at some "fair" reaction. Several studies concentrate on intraday stock price overreactions and reversals. In this Journal of Advanced Studies in Finance study, I make an effort to "move one step forward" and ask the following question: "May we expect consistent (overnight) stock price reversals or continuations between consecutive trading days?" Or in other words: "Will opening stock returns be higher or lower following the days that were closed with price increases or decreases?" Table 1. Descriptive statistics of sample stocks' intraday and daily returns and price differences Company (Ticker symbol) Alcoa Inc. (AA) American Express (AXP) Boeing (BA) Bank of America (BAC) Caterpillar (CAT) Cisco Systems (CSCO) Chevron Corporation (CVX) E.I. Du Pont de Nemours (DD) Walt Disney (DIS) General Electric (GE) Home Depot Inc. (HD) Hewlett-Packard (HPQ) IBM (IBM) Intel Corporation (INTC) Johnson & Johnson (JNJ) JP Morgan Chase & Co (JPM) Kraft Foods Inc. (KFT) Coca-Cola (KO) McDonald's Corporation (MCD) 3M Company (MMM) Merck & Company Inc. (MRK) Microsoft Corporation (MSFT) Pfizer Inc. (PFE) Procter & Gamble (PG) AT&T Inc. (T) The Travelers Companies (TRV) United Technologies Corp. (UTX) Verizon Communications (VZ) Wal-Mart Stores Inc. (WMT) Exxon Mobil Corporation (XOM) Opening return, % Mean St. Dev. 0.158 1.551 -0.008 1.212 0.033 1.004 0.104 2.045 0.086 1.198 0.026 1.410 0.032 0.808 0.039 0.879 -0.048 1.094 0.086 1.275 0.001 1.047 -0.116 1.386 -0.061 0.956 0.038 1.377 0.005 0.715 0.043 1.436 -0.013 0.829 -0.009 0.663 0.007 0.860 0.017 0.756 -0.022 1.110 0.017 1.020 0.054 1.047 -0.043 0.625 0.053 0.933 0.042 0.996 0.042 0.824 0.035 0.839 0.018 0.735 -0.005 0.801 Daily return, % Mean St. Dev. -0.004 2.884 0.054 2.621 0.047 1.979 0.010 3.568 0.076 2.189 0.019 2.365 0.058 1.762 0.029 1.867 0.038 2.020 -0.005 2.120 0.011 2.020 0.031 2.225 0.033 1.597 0.017 2.338 0.021 1.225 0.047 2.936 0.020 1.393 0.033 1.304 0.071 1.566 0.029 1.512 0.012 1.919 0.015 1.889 -0.004 1.693 0.035 1.183 0.022 1.753 0.037 2.088 0.054 1.708 0.023 1.681 0.011 1.391 0.048 1.710 High-to-close, % Mean St. Dev. 1.695 1.882 1.400 1.758 1.225 1.170 1.582 2.582 1.317 1.362 1.453 1.416 1.039 1.127 1.172 1.220 1.163 1.163 1.226 1.436 1.315 1.279 1.303 1.324 0.943 0.983 1.430 1.366 0.771 0.751 1.513 1.934 0.903 1.043 0.808 0.849 1.003 1.008 0.937 0.948 1.151 1.176 1.147 1.148 1.120 1.066 0.744 0.770 1.172 1.170 1.241 1.647 1.022 1.040 1.065 1.049 0.950 0.868 0.989 1.084 Low-to-close, % Mean St. Dev. 1.653 1.693 1.508 1.964 1.303 1.239 1.469 2.600 1.408 1.365 1.374 1.363 1.107 1.158 1.209 1.221 1.320 1.363 1.234 1.545 1.326 1.416 1.423 1.401 1.001 1.016 1.389 1.410 0.794 0.848 1.566 2.115 0.978 1.143 0.878 0.839 1.072 1.060 1.986 1.054 1.204 1.233 1.095 1.182 1.088 1.140 0.867 1.394 1.178 1.218 1.250 1.516 1.099 1.137 1.147 1.188 0.934 0.982 1.075 1.124 I employ the high-to-close stock price differences as a proxy for the "end-of-the-day" stock price decreases, and the low-to-close stock price differences as a proxy for the "end-of-the-day" stock price increases. For each of the sample stocks and for each of the trading days, I compare the respective price differences, and test the following hypotheses with respect to the next day's opening session: Hypothesis 1a (overnight price reversals following price increases): Opening stock price returns should be lower following the days with prevailing low-to-close price differences3. against (overnight price continuations following price increases): Opening stock price returns should be higher following the days with prevailing low-to-close price differences. Hypothesis 1b (overnight price reversals following price decreases): Opening stock price returns should be higher following the days with prevailing high-to-close price differences4. against (overnight price continuations following price decreases): Opening stock price returns should be lower following the days with prevailing high-to-close price differences. Table 2 provides the hypotheses tests by comparing the means of opening stock returns ( RO,it ) following the days with prevailing high-to-close ( RHC,it 1 ) and prevailing low-to-close ( RHC,it 1 ) price differences. The results clearly support the 'reversal' versions of the hypotheses. For 28 out of 30 sample Days when for the respective stock, the high-to-close price difference was smaller than the low-to-close price difference. Days when for the respective stock, the high-to-close price difference was greater or equal to the low-to-close price difference. 165 stocks, mean opening returns are higher following the days with prevailing high-to-close price differences, that is, following the days closed with price decreases. 17 of the respective mean opening return differences are statistically significant, including 13 at the 5% level, and 11 at the 1% level. Moreover, for 29 out of 30 stocks, mean opening returns are positive if on the previous trading day, high-to-close price difference prevailed, and for 19 out of 30 stocks, mean opening returns are negative if on the previous trading day, low-to-close price difference prevailed. Thus, the results demonstrate that opening stock prices display a reverting behavior following previous day's "end-of-the-day" price moves. Such behavior may be regarded as "overnight reversals", and seems to contradict the market efficiency. 3.2. Portfolios based on "overnight reversals" stock price behavior Previous Subsection shows that stock i's opening return on day t tends to be higher if RHC,it holds. This evidence suggests that there are overnight reversals of previous day's "end-of-the-day" price moves, or in other words, that the 'reversal' versions of Hypotheses 1a and 1b jointly hold. Now, I proceed to separately testing the Hypotheses. I do that by constructing a number of portfolios, approach that in addition to testing the Hypotheses, allows me to consider a number of potentially profitable investment strategies. All the portfolios are built of the sample stocks and upon the idea of holding during the opening sessions and daily adjusting a long position in the stocks that according to the "overnight reversals" pattern are expected to yield high opening returns, that is, the stocks that on the previous trading day showed relatively large high-to-close price differences, and a short position in the stocks that according to "overnight reversals" pattern are expected to yield low opening returns, that is, the stocks that on the previous trading day showed relatively large low-to-close price differences. To get a proxy for the "relatively large" high-to-close and low-to-close price differences, for each of the stocks and for each of the trading days, I compare these measures to each other, as in the previous Subsection, and also to the respective mean and median measures for the total sample on the respective trading day. The respective positions in the portfolios are taken at the end of each trading day and closed at the end of the next day's opening session. The total values of the long and the short positions are supposed to be equal, that is, the total market value of each portfolio after the positions are taken is supposed to be zero5. a) Portfolios based on low-to-close price differences: Portfolio LA ("Low-to-close price difference compared to the sample Average"): Portfolio implying an equally-weighted long position for a day's opening session in the stocks whose previous day's low-to-close price differences were smaller than the sample average, and an equally-weighted short position for the day's opening session in the rest of the sample stocks. Portfolio LM ("Low-to-close price difference compared to the sample Median"): Portfolio implying an equally-weighted long position for a day's opening session in the stocks whose previous day's low-to-close price differences were smaller than the sample median, and an equally-weighted short position for the day's opening session in the rest of the sample stocks. Table 2. Opening stock returns following the days with prevailing high-to-close and low-to-close price differences Company (Ticker symbol) Mean opening returns ( RO,it ), %, for the days when: RHC,it Alcoa Inc. (AA) American Express (AXP) Boeing (BA) Bank of America (BAC) Caterpillar (CAT) Cisco Systems (CSCO) Chevron Corporation (CVX) RHC,it (No. of days) 0.136 (1234) 0.032 (1179) 0.053 (1171) 0.051 (1256) 0.096 (1168) 0.075 (1257) 0.064 (1173) (No. of days) 0.179 (1220) -0.045 (1275) 0.015 (1283) 0.160 (1198) 0.077 (1286) -0.027 (1197) 0.001 (1281) Difference (t-statistic) -0.043 (-0.69) 0.077 (1.56) 0.038 (0.93) -0.109 (-1.32) 0.19 (0.39) *0.102 (1.80) *0.063 (1.86) In constructing the portfolios, I assume that all the long and the short positions are taken at the closing prices based on the high-to-close and the low-to-close individual stocks', mean and median measures that are readily available. Of course, at the time of the transactions, the closing stock prices, and respectively, the high-to-close and the low-to-close price differences, are yet not exactly known. But I suggest that if the transactions are performed sufficiently close to the market closing time, then the actual high-to-close and the low-to-close price differences and the transaction prices should be sufficiently close to those based on the closing prices. 166 Journal of Advanced Studies in Finance E.I. Du Pont de Nemours(DD) 0.095 (1189) Walt Disney (DIS) 0.015 (1133) General Electric (GE) 0.127 (1224) Home Depot Inc. (HD) 0.040 (1232) Hewlett-Packard (HPQ) -0.074 (1125) IBM (IBM) 0.001 (1167) Intel Corporation (INTC) 0.072 (1219) Johnson & Johnson (JNJ) 0.028 (1235) JP Morgan Chase & Co(JPM) 0.059 (1184) Kraft Foods Inc. (KFT) 0.040 (1191) Coca-Cola (KO) 0.045 (1118) McDonald's Corporation (MCD) 0.057 (1149) 3M Company (MMM) 0.049 (1196) Merck & Company Inc. (MRK) 0.012 (1176) Microsoft Corporation (MSFT) 0.079 (1278) Pfizer Inc. (PFE) 0.131 (1258) Procter & Gamble (PG) 0.015 (1105) AT&T Inc. (T) 0.089 (1235) The Travelers Companies(TRV) 0.058 (1209) United Technologies Corp. (UTX) 0.064 (1159) Verizon Communications (VZ) 0.075 (1196) Wal-Mart Stores Inc. (WMT) 0.077 (1233) Exxon Mobil Corporation (XOM) 0.050 (1113) Asterisks denote two-tailed p-values: *p<0.10; **p<0.05; ***p<0.01. -0.014 (1265) -0.104 (1321) 0.046 (1230) -0.037 (1222) -0.151 (1329) -0.118 (1287) 0.001 (1235) -0.017 (1219) 0.027 (1270) -0.062 (1263) -0.055 (1336) -0.038 (1305) -0.013 (1258) -0.054 (1278) -0.050 (1176) -0.027 (1196) -0.089 (1349) 0.016 (1219) 0.027 (1245) 0.022 (1295) -0.001 (1258) -0.041 (1221) -0.051 (1341) ***0.109 (3.05) ***0.119 (2.69) 0.081 (1.56) *0.077 (1.82) 0.077 (1.38) ***0.119 (3.10) 0.071 (1.23) 0.045 (1.54) 0.032 (0.56) ***0.102 (3.06) ***0.100 (3.75) ***0.095 (2.72) **0.062 (2.04) 0.066 (1.49) ***0.129 (3.14) ***0.158 (3.74) ***0.104 (4.12) *0.073 (1.93) 0.031 (0.76) 0.042 (1.25) **0.076 (2.30) ***0.118 (3.97) ***0.101 (3.09) The idea behind these three portfolios is based on exploiting the 'reversal' version of Hypothesis 1a. That is, an investor is supposed to hold during the opening sessions an equally-weighted long position in the stocks whose previous day's low-to-close price differences were relatively small, and an equally-weighted short position in the stocks whose previous day's low-to-close price differences were relatively large. The portfolios do not suggest any initial investment, since the total values of the long and the short positions are equal. b) Portfolios based on high-to-close price differences: Portfolio HA ("High-to-close price difference compared to the sample Average"): Portfolio implying an equally-weighted long position for a day's opening session in the stocks whose previous day's high-to-close price differences were greater than the sample average, and an equally-weighted short position for the day's opening session in the rest of the sample stocks. Portfolio HM ("High-to-close price difference compared to the sample Median"): Portfolio implying an equally-weighted long position for a day's opening session in the stocks whose previous day's high-to-close price differences were greater than the sample median, and an equally-weighted short position for the day's opening session in the rest of the sample stocks. The idea behind these three portfolios is based on exploiting the 'reversal' version of Hypothesis 1b. That is, an investor is supposed to hold during the opening sessions an equally-weighted long position in the stocks whose previous day's high-to-close price differences were relatively large, and an equally-weighted short position in the stocks whose previous day's high-to-close price differences were relatively small. The portfolios do not suggest any initial investment, since the total values of the long and the short positions are equal. c) Portfolio based on comparison of low-to-close and high-to-close price differences: Portfolio L-H ("Low-to-close ­ High-to-close price differences"): Portfolio implying an equally-weighted long position for a day's opening session in the stocks whose previous day's high-to-close price differences were greater or equal to their previous day's low-to-close price differences, and an equally-weighted short position for the day's opening session in the rest of the sample stocks. The idea of this portfolio is based on jointly exploiting the 'reversal' versions of Hypotheses 1a and 1b. This portfolio too, does not suggest any initial investment, since the total values of the long and the short positions are equal. According to the definitions, the portfolios in paragraphs (a) and (b) refer to testing Hypotheses 1a and 1b, respectively, while the portfolio in paragraph (c) tests for the joint effect of both tendencies, the issue presented and discussed in Table 2 in the previous Subsection. Table 3 concentrates the daily performance measures over the sampling period for all the five portfolios. Strikingly, all the portfolios yield significantly positive mean daily returns. These results, first of all, provide a strong support for both research hypotheses, even when these are separately tested. That is, opening stock returns are significantly higher following the days with relatively large high-to-close price differences (with respect to the sample's average or median) and also following the days of the relatively small low-to-close price differences. Moreover, from the practical point of view, at least if the trading commissions are not a problem, the five portfolios represent potentially profitable investment strategies. The mean opening returns of about 0.1 percentage point (or even smaller) may, at the first glance, seem not quite impressive, but since we are talking about single-day opening returns, the mean annual return of about 25% on Portfolio L-H or about 30% on portfolio HA, for example, look promising (recall that the portfolios do not request any initial investments and yield significantly positive returns). Table 3. Historical performance measures of the portfolios based on the "overnight reversals" stock price behavior Daily-adjusted Portfolio performance measures (opening returns) over the sampling period (2456 days) portfolios Mean, % Standard Deviation, % t-statistic (Mean=0) Portfolio LA 0.040 0.478 ***4.10 Portfolio LM 0.045 0.406 ***5.50 Portfolio HA 0.104 0.466 ***11.09 Portfolio HM 0.091 0.388 ***11.60 Portfolio L-H 0.090 0.541 ***8.23 Asterisks denote two-tailed p-values: *p<0.10; **p<0.05; ***p<0.01. Overall, the results in this Section strongly indicate that the "end-of-the-day" stock price moves tend to be reversed in the next day's opening session. Investment strategies built upon the expectation of "overnight reversals" may, therefore, possess a non-negligible potential. Conclusion The main goal of the present study is to shed light on the dynamics of stock price reversals and namely, on the overnight reversals. I expect that following some relatively large price moves towards the end of the trading days, there may be price reversals during the next days' opening sessions. To account for price changes at the end of the trading days, I employ high-to-close and low-to-close price differences. I employ intraday price data on thirty stocks currently making up the Dow Jones Industrial Index, and find supporting evidence for my research hypothesis. For each trading day, I compare each stock's high-to-close and low-to-close price changes, and also compare them to the same day's average and median changes for the total sample of stocks, and document that opening returns tend to be higher following the days with relatively large high-to-close price changes (price decreases at the end of the day), and lower following the days with relatively large low-to-close price changes (price increases at the end of the day). These findings imply that stock price changes towards the end of a trading day may contain an element of overreaction to be reversed right at the beginning of the next trading day. Such "overnight reversals" price behavior seems to contradict the market efficiency. Furthermore, I test if on the basis of these findings it is possible to define potentially profitable investment strategies. I construct a number of portfolios based on the opening trading sessions and involving a long position in the stocks on the days when, according to the findings, their opening returns are expected to be high and a short position in the stocks on the days when, according to the findings, their opening returns are expected to be low. All the portfolios are found to yield significantly positive returns, providing an evidence for the practical applicability of the "overnight reversals" pattern in stock prices. Overall, my findings amplify the results documented in the previous literature with respect to the profit potential embedded in the short-term stock price reversals. Lehmann (1990); Jegadeesh (1990), and Conrad et al. (1994) analyze stock price reactions to news and conclude that these are usually too strong (overreaction), and therefore, short-term price reversals may be generally expected. Zawadowski et al. (2006), and Grant et al. (2005) concentrate on especially short time intervals, and document intraday stock price reversals. In this study, I make an effort to "move one step forward" and show that there are consistent (overnight) stock price reversals between consecutive trading days, or in other words, that opening stock returns tend to be higher following the days that were closed with price decreases and lower following the days that were closed with price increases. Journal of Advanced Studies in Finance To summarize, at least in a perfect stock market with no commissions, the daily-adjusted strategies based on the expectations of the "overnight reversals" look promising. This may prove a valuable result for both financial theoreticians in their eternal discussion about stock market efficiency, and practitioners in search of potentially profitable investment strategies. Potential directions for further research may include expending the analysis to other stock exchanges and greater samples, though in the latter case some care has to be taken when defining the comparative benchmarks for high-to-close and low-to-close price difference measures, and also applying similar kind of analysis to longer time intervals. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Advanced Studies in Finance de Gruyter

Overnight Stock Price Reversals

Loading next page...
 
/lp/de-gruyter/overnight-stock-price-reversals-hyBhNr19Rz
Publisher
de Gruyter
Copyright
Copyright © 2012 by the
ISSN
2068-8393
DOI
10.2478/v10259-012-0011-1
Publisher site
See Article on Publisher Site

Abstract

In present study, I explore the dynamics of stock price reversals. In particular, I try to shed light on the overnight reversals, that is, on the price reversals between the end of a trading day and the opening session of the next trading day. To account for the "end-of-the-day" price moves, for each of the stocks currently making up the Dow Jones Industrial Index, I compare, on the daily basis, the high-to-close and the low-to-close price changes, and also compare them to the same day's average and median changes for the total sample of stocks. I document that opening returns tend to be higher following the days with relatively large high-to-close price changes (price decreases at the end of the day), and lower following the days with relatively large low-to-close price changes (price increases at the end of the day). Such "overnight reversals" price behavior seems to contradict the market efficiency. Finally, I construct five portfolios based on the opening trading sessions and involving a long position in the stocks on the days when, according to the "overnight reversals" behavior, their opening returns are expected to be high and a short position in the stocks on the days when their opening returns are expected to be low. All the portfolios are found to yield significantly positive returns, providing an evidence for the practical applicability of the "overnight reversals" pattern in stock prices. Keywords: Intraday Stock Prices; Opening Stock Returns; Overreaction; Stock Price Reversals JEL Classifications: G11, G14, G19. 1. Introduction In informationally efficient asset markets, prices incorporate news quickly and accurately and investors cannot predict future returns and make abnormal profits. However, the empirical results of Shiller (1984), and DeBondt, and Thaler (1985) seriously challenge the notion of efficient capital markets and indicate that abnormal profits are possible using historical information. More specifically, they demonstrate that a `contrarian' strategy of going long a portfolio of extreme prior losers and going short a portfolio of extreme prior winners will produce long-term abnormal profits. This finding is explained the tendency of investors to overreact to information, resulting in subsequent price reversals. Since these pioneering studies, a large volume of theoretical and empirical research work has analyzed price overreaction in financial markets. A number of potential explanations for this phenomenon have been suggested, the most popular of them including bid-ask biases, investor psychology, multifactor pricing models, size and transaction costs. For example, short-term reversals may be induced by prices bouncing between bidask quotes. Jegadeesh, and Titman (1995) show that inventory imbalances may cause negative short - term serial correlation in prices, while Roll (1984) shows that due to the dealers order processing costs the bid-ask bounce may explain short - term negative serial correlation (for bid - ask explanations, see also, Cox, and Peterson 1994, Atkins, and Dyl 1990; Park 1995). Fama, and French (1996) find that long-term equity return reversals can be explained within the context of a multifactor asset pricing model, while Zarowin (1990) finds that contrarian profits may be due to a size effect in stock returns. Other authors attempt to explain overreaction by means of investor psychology (e.g., Barberis et al. 1998; Daniel et al. 1998; Lakonishok et al. 1994; Hong, and Stein 1999; Odean 1998), and present different channels through which investor psychology can lead to inefficiencies in securities' returns. The rationale for these studies originates from evidence of empirical psychology that individuals tend at times to underreact or overreact (Kahneman, and Tversky 1982; Griffin, and Tversky 1992). For example, the confidence model provided by Daniel et al. (1998) predicts long-run negative serial correlation, and the model developed by Barberis et al. (1998) argues that investors overreact to strong and salient information. The latter model is based on two well - Journal of Advanced Studies in Finance established human psychological characteristics: representativeness and conservatism, and suggests that overreaction is expected to cause future reversals as prices revert to their fundamental value. In order to distinguish stock price overreaction and market inefficiency from predictable changes in expected returns, Lehmann (1990) suggests examining returns over short time intervals. In fact, the focus on long-term dynamics in stock returns in the papers by Shiller (1984), and De Bondt, and Thaler (1985) is more recently realigned to short - run return behavior, ranging over time periods from a few days up to a month, in the major part of the subsequent literature (e.g., Zarowin 1989; Atkins, and Dyl 1990; Cox, and Peterson 1994; Park 1995; Bowman, and Iverson 1998; Nam et al. 2001). The major focus of these studies is on identifying potentially profitable contrarian strategies built on a reverting behavior of stock prices in the short run. For example, Lehmann (1990), and Jegadeesh (1990) show that contrarian strategies that exploit the short-run return reversals in individual stocks generate abnormal returns of about 1.7% per week and 2.5% per month, respectively. Remarkably, Conrad et al. (1994) document that reversal profitability increases with trading activity. A continuously growing body of literature concentrates on even shorter time intervals, and studies intraday price reversals. Zawadowski et al. (2006) analyze a large sample of NASDAQ-traded stocks, and define "large intraday price changes" as intraday price changes bigger than a certain level of 2-6% price change within 10-120 minutes, and alternatively, as intraday price changes exceeding 6-10 times the normal volatility during that time of the day. They document significant stock price reversals following these initial changes, implying that the latter contain at least some element of overreaction. Grant et al. (2005) look at large changes in the prices of S&P 500 futures during the opening trading session (with respect to the previous day's closing price), and find that such changes are followed by significant price reversals in the first half-an-hour to hour after the market opening. The main goal of the present study is to shed a little more light on the dynamics of stock price reversals and namely, on the overnight reversals. The study is logically connected to the study by Grant et al. (2005), but looks at the "opposite" order of events and analyzes a more general case. Namely, I expect that following some relatively large price moves towards the end of the trading days, there may be price reversals during the next days' opening sessions1. To account for price changes at the end of the trading days, I employ high-to-close and low-to-close price differences. The use of these measures is in line with Becker et al. (2008), and Klossner et al. (2012), who employ them as integral parts of their intraday upside and downside volatility measures, respectively. The focus on the next days' opening sessions as potential "reversal periods" is motivated by the short-time nature of the reversals, in general, and by the observation (e.g., Harris 1986) that the predominant portion of stock price moves takes place within the first minutes of trading. I analyze intraday price data on thirty stocks currently making up the Dow Jones Industrial Index, and find supporting evidence for my research hypothesis. For each trading day, I compare each stock's high-to-close and low-to-close price changes, and also compare them to the same day's average and median changes for the total sample of stocks, and document that opening returns tend to be higher following the days with relatively large high-to-close price changes (price decreases at the end of the day), and lower following the days with relatively large low-to-close price changes (price increases at the end of the day). These findings imply that stock price changes towards the end of a trading day may contain an element of overreaction to be reversed right at the beginning of the next trading day. Based on these findings, for the opening trading sessions, I construct a number of daily-adjusted portfolios involving a long position in the opening session in the stocks on the days when, according to the findings, their opening returns are expected to be high and a short position in the opening session in the stocks on the days when, according to the findings, their opening returns are expected to be low, and demonstrate that the (opening) returns on these portfolios are significantly positive. The rest of the paper is structured as follows: In Section 2, I describe the data sample. Section 3 comprises the research hypotheses and the results. Section 4 concludes. 2. Data description For the purposes of present research, I employ daily opening, high, low and closing prices of thirty stocks currently making up the Dow Jones Industrial Index over the period comprised from January 2, 2002 to September 30, 2011 (overall, 2456 trading days).2 I adjust all the prices to dividend payments and stock splits, and for each stock in the sample and for each trading day in the sampling period, calculate: I test this hypothesis against the opposite one, suggesting that in the opening sessions, stock prices may continue to move in the direction of the respective changes that took place towards the end of the previous days. 2 The data were taken from the Yahoo Finance website. 1. Stock's opening return (i.e., stock price's change from last day's closing price to today's opening price) as: RO,it PO,it PC ,it (1) where: RO,it is stock i's opening return on day t; PO ,it is stock i's opening price on day t; and PC ,it is stock i's closing price on day t-1. 2.Stock's daily return (i.e., stock price's change from last day's closing price to today's closing price), as: RD ,it PC ,it PC ,it (2) where: R D ,it is stock i's daily return on day t. 3.Stock's high-to-close price difference, as: R HC ,it PH ,it PC ,it (3) where: RHC,it is stock i's high-to-close price difference on day t; and PH ,it is stock i's highest price on day t. 4.Stock's low-to-close price difference, as: RLC ,it PC ,it PL ,it (4) where: is stock i's low-to-close price difference on day t; and PL,it is stock i's lowest price on day t. Of course, since the following relations between the intraday prices hold: PH ,it PC ,it ; and PL,it PC ,it (5) the last two price differences are defined so that they are non-negative, representing the absolute values of the respective price changes. Table 1 comprises the basic descriptive statistics of the intraday price differences and returns for the thirty sample stocks. At this stage, we may note that, as it might be expected for the largest industrial companies of the US, 27 out of 30 stocks have positive mean daily returns, the remaining 3 showing negative, yet close to zero daily returns. Overall, the mean daily returns range from -0.005 to 0.076 percentage points, with standard deviations ranging from 1.183 to 3.568 percentage points. The two intraday mean price differences are highly correlated in the cross-section, that is, all of them are relatively high for certain stocks and relatively low for other ones. One more thing to note is that for 23 out of 30 stocks, the mean low-to-close differences are greater than the mean high-to-close differences. 3. Research hypotheses and results 3.1. Effect of the "end-of-the-day" stock price moves on the next day's opening returns The concept of stock price reversals is well-documented and widely-discussed in financial literature. Many studies show that stock prices often overreact to news and subsequently revert themselves in order to arrive at some "fair" reaction. Several studies concentrate on intraday stock price overreactions and reversals. In this Journal of Advanced Studies in Finance study, I make an effort to "move one step forward" and ask the following question: "May we expect consistent (overnight) stock price reversals or continuations between consecutive trading days?" Or in other words: "Will opening stock returns be higher or lower following the days that were closed with price increases or decreases?" Table 1. Descriptive statistics of sample stocks' intraday and daily returns and price differences Company (Ticker symbol) Alcoa Inc. (AA) American Express (AXP) Boeing (BA) Bank of America (BAC) Caterpillar (CAT) Cisco Systems (CSCO) Chevron Corporation (CVX) E.I. Du Pont de Nemours (DD) Walt Disney (DIS) General Electric (GE) Home Depot Inc. (HD) Hewlett-Packard (HPQ) IBM (IBM) Intel Corporation (INTC) Johnson & Johnson (JNJ) JP Morgan Chase & Co (JPM) Kraft Foods Inc. (KFT) Coca-Cola (KO) McDonald's Corporation (MCD) 3M Company (MMM) Merck & Company Inc. (MRK) Microsoft Corporation (MSFT) Pfizer Inc. (PFE) Procter & Gamble (PG) AT&T Inc. (T) The Travelers Companies (TRV) United Technologies Corp. (UTX) Verizon Communications (VZ) Wal-Mart Stores Inc. (WMT) Exxon Mobil Corporation (XOM) Opening return, % Mean St. Dev. 0.158 1.551 -0.008 1.212 0.033 1.004 0.104 2.045 0.086 1.198 0.026 1.410 0.032 0.808 0.039 0.879 -0.048 1.094 0.086 1.275 0.001 1.047 -0.116 1.386 -0.061 0.956 0.038 1.377 0.005 0.715 0.043 1.436 -0.013 0.829 -0.009 0.663 0.007 0.860 0.017 0.756 -0.022 1.110 0.017 1.020 0.054 1.047 -0.043 0.625 0.053 0.933 0.042 0.996 0.042 0.824 0.035 0.839 0.018 0.735 -0.005 0.801 Daily return, % Mean St. Dev. -0.004 2.884 0.054 2.621 0.047 1.979 0.010 3.568 0.076 2.189 0.019 2.365 0.058 1.762 0.029 1.867 0.038 2.020 -0.005 2.120 0.011 2.020 0.031 2.225 0.033 1.597 0.017 2.338 0.021 1.225 0.047 2.936 0.020 1.393 0.033 1.304 0.071 1.566 0.029 1.512 0.012 1.919 0.015 1.889 -0.004 1.693 0.035 1.183 0.022 1.753 0.037 2.088 0.054 1.708 0.023 1.681 0.011 1.391 0.048 1.710 High-to-close, % Mean St. Dev. 1.695 1.882 1.400 1.758 1.225 1.170 1.582 2.582 1.317 1.362 1.453 1.416 1.039 1.127 1.172 1.220 1.163 1.163 1.226 1.436 1.315 1.279 1.303 1.324 0.943 0.983 1.430 1.366 0.771 0.751 1.513 1.934 0.903 1.043 0.808 0.849 1.003 1.008 0.937 0.948 1.151 1.176 1.147 1.148 1.120 1.066 0.744 0.770 1.172 1.170 1.241 1.647 1.022 1.040 1.065 1.049 0.950 0.868 0.989 1.084 Low-to-close, % Mean St. Dev. 1.653 1.693 1.508 1.964 1.303 1.239 1.469 2.600 1.408 1.365 1.374 1.363 1.107 1.158 1.209 1.221 1.320 1.363 1.234 1.545 1.326 1.416 1.423 1.401 1.001 1.016 1.389 1.410 0.794 0.848 1.566 2.115 0.978 1.143 0.878 0.839 1.072 1.060 1.986 1.054 1.204 1.233 1.095 1.182 1.088 1.140 0.867 1.394 1.178 1.218 1.250 1.516 1.099 1.137 1.147 1.188 0.934 0.982 1.075 1.124 I employ the high-to-close stock price differences as a proxy for the "end-of-the-day" stock price decreases, and the low-to-close stock price differences as a proxy for the "end-of-the-day" stock price increases. For each of the sample stocks and for each of the trading days, I compare the respective price differences, and test the following hypotheses with respect to the next day's opening session: Hypothesis 1a (overnight price reversals following price increases): Opening stock price returns should be lower following the days with prevailing low-to-close price differences3. against (overnight price continuations following price increases): Opening stock price returns should be higher following the days with prevailing low-to-close price differences. Hypothesis 1b (overnight price reversals following price decreases): Opening stock price returns should be higher following the days with prevailing high-to-close price differences4. against (overnight price continuations following price decreases): Opening stock price returns should be lower following the days with prevailing high-to-close price differences. Table 2 provides the hypotheses tests by comparing the means of opening stock returns ( RO,it ) following the days with prevailing high-to-close ( RHC,it 1 ) and prevailing low-to-close ( RHC,it 1 ) price differences. The results clearly support the 'reversal' versions of the hypotheses. For 28 out of 30 sample Days when for the respective stock, the high-to-close price difference was smaller than the low-to-close price difference. Days when for the respective stock, the high-to-close price difference was greater or equal to the low-to-close price difference. 165 stocks, mean opening returns are higher following the days with prevailing high-to-close price differences, that is, following the days closed with price decreases. 17 of the respective mean opening return differences are statistically significant, including 13 at the 5% level, and 11 at the 1% level. Moreover, for 29 out of 30 stocks, mean opening returns are positive if on the previous trading day, high-to-close price difference prevailed, and for 19 out of 30 stocks, mean opening returns are negative if on the previous trading day, low-to-close price difference prevailed. Thus, the results demonstrate that opening stock prices display a reverting behavior following previous day's "end-of-the-day" price moves. Such behavior may be regarded as "overnight reversals", and seems to contradict the market efficiency. 3.2. Portfolios based on "overnight reversals" stock price behavior Previous Subsection shows that stock i's opening return on day t tends to be higher if RHC,it holds. This evidence suggests that there are overnight reversals of previous day's "end-of-the-day" price moves, or in other words, that the 'reversal' versions of Hypotheses 1a and 1b jointly hold. Now, I proceed to separately testing the Hypotheses. I do that by constructing a number of portfolios, approach that in addition to testing the Hypotheses, allows me to consider a number of potentially profitable investment strategies. All the portfolios are built of the sample stocks and upon the idea of holding during the opening sessions and daily adjusting a long position in the stocks that according to the "overnight reversals" pattern are expected to yield high opening returns, that is, the stocks that on the previous trading day showed relatively large high-to-close price differences, and a short position in the stocks that according to "overnight reversals" pattern are expected to yield low opening returns, that is, the stocks that on the previous trading day showed relatively large low-to-close price differences. To get a proxy for the "relatively large" high-to-close and low-to-close price differences, for each of the stocks and for each of the trading days, I compare these measures to each other, as in the previous Subsection, and also to the respective mean and median measures for the total sample on the respective trading day. The respective positions in the portfolios are taken at the end of each trading day and closed at the end of the next day's opening session. The total values of the long and the short positions are supposed to be equal, that is, the total market value of each portfolio after the positions are taken is supposed to be zero5. a) Portfolios based on low-to-close price differences: Portfolio LA ("Low-to-close price difference compared to the sample Average"): Portfolio implying an equally-weighted long position for a day's opening session in the stocks whose previous day's low-to-close price differences were smaller than the sample average, and an equally-weighted short position for the day's opening session in the rest of the sample stocks. Portfolio LM ("Low-to-close price difference compared to the sample Median"): Portfolio implying an equally-weighted long position for a day's opening session in the stocks whose previous day's low-to-close price differences were smaller than the sample median, and an equally-weighted short position for the day's opening session in the rest of the sample stocks. Table 2. Opening stock returns following the days with prevailing high-to-close and low-to-close price differences Company (Ticker symbol) Mean opening returns ( RO,it ), %, for the days when: RHC,it Alcoa Inc. (AA) American Express (AXP) Boeing (BA) Bank of America (BAC) Caterpillar (CAT) Cisco Systems (CSCO) Chevron Corporation (CVX) RHC,it (No. of days) 0.136 (1234) 0.032 (1179) 0.053 (1171) 0.051 (1256) 0.096 (1168) 0.075 (1257) 0.064 (1173) (No. of days) 0.179 (1220) -0.045 (1275) 0.015 (1283) 0.160 (1198) 0.077 (1286) -0.027 (1197) 0.001 (1281) Difference (t-statistic) -0.043 (-0.69) 0.077 (1.56) 0.038 (0.93) -0.109 (-1.32) 0.19 (0.39) *0.102 (1.80) *0.063 (1.86) In constructing the portfolios, I assume that all the long and the short positions are taken at the closing prices based on the high-to-close and the low-to-close individual stocks', mean and median measures that are readily available. Of course, at the time of the transactions, the closing stock prices, and respectively, the high-to-close and the low-to-close price differences, are yet not exactly known. But I suggest that if the transactions are performed sufficiently close to the market closing time, then the actual high-to-close and the low-to-close price differences and the transaction prices should be sufficiently close to those based on the closing prices. 166 Journal of Advanced Studies in Finance E.I. Du Pont de Nemours(DD) 0.095 (1189) Walt Disney (DIS) 0.015 (1133) General Electric (GE) 0.127 (1224) Home Depot Inc. (HD) 0.040 (1232) Hewlett-Packard (HPQ) -0.074 (1125) IBM (IBM) 0.001 (1167) Intel Corporation (INTC) 0.072 (1219) Johnson & Johnson (JNJ) 0.028 (1235) JP Morgan Chase & Co(JPM) 0.059 (1184) Kraft Foods Inc. (KFT) 0.040 (1191) Coca-Cola (KO) 0.045 (1118) McDonald's Corporation (MCD) 0.057 (1149) 3M Company (MMM) 0.049 (1196) Merck & Company Inc. (MRK) 0.012 (1176) Microsoft Corporation (MSFT) 0.079 (1278) Pfizer Inc. (PFE) 0.131 (1258) Procter & Gamble (PG) 0.015 (1105) AT&T Inc. (T) 0.089 (1235) The Travelers Companies(TRV) 0.058 (1209) United Technologies Corp. (UTX) 0.064 (1159) Verizon Communications (VZ) 0.075 (1196) Wal-Mart Stores Inc. (WMT) 0.077 (1233) Exxon Mobil Corporation (XOM) 0.050 (1113) Asterisks denote two-tailed p-values: *p<0.10; **p<0.05; ***p<0.01. -0.014 (1265) -0.104 (1321) 0.046 (1230) -0.037 (1222) -0.151 (1329) -0.118 (1287) 0.001 (1235) -0.017 (1219) 0.027 (1270) -0.062 (1263) -0.055 (1336) -0.038 (1305) -0.013 (1258) -0.054 (1278) -0.050 (1176) -0.027 (1196) -0.089 (1349) 0.016 (1219) 0.027 (1245) 0.022 (1295) -0.001 (1258) -0.041 (1221) -0.051 (1341) ***0.109 (3.05) ***0.119 (2.69) 0.081 (1.56) *0.077 (1.82) 0.077 (1.38) ***0.119 (3.10) 0.071 (1.23) 0.045 (1.54) 0.032 (0.56) ***0.102 (3.06) ***0.100 (3.75) ***0.095 (2.72) **0.062 (2.04) 0.066 (1.49) ***0.129 (3.14) ***0.158 (3.74) ***0.104 (4.12) *0.073 (1.93) 0.031 (0.76) 0.042 (1.25) **0.076 (2.30) ***0.118 (3.97) ***0.101 (3.09) The idea behind these three portfolios is based on exploiting the 'reversal' version of Hypothesis 1a. That is, an investor is supposed to hold during the opening sessions an equally-weighted long position in the stocks whose previous day's low-to-close price differences were relatively small, and an equally-weighted short position in the stocks whose previous day's low-to-close price differences were relatively large. The portfolios do not suggest any initial investment, since the total values of the long and the short positions are equal. b) Portfolios based on high-to-close price differences: Portfolio HA ("High-to-close price difference compared to the sample Average"): Portfolio implying an equally-weighted long position for a day's opening session in the stocks whose previous day's high-to-close price differences were greater than the sample average, and an equally-weighted short position for the day's opening session in the rest of the sample stocks. Portfolio HM ("High-to-close price difference compared to the sample Median"): Portfolio implying an equally-weighted long position for a day's opening session in the stocks whose previous day's high-to-close price differences were greater than the sample median, and an equally-weighted short position for the day's opening session in the rest of the sample stocks. The idea behind these three portfolios is based on exploiting the 'reversal' version of Hypothesis 1b. That is, an investor is supposed to hold during the opening sessions an equally-weighted long position in the stocks whose previous day's high-to-close price differences were relatively large, and an equally-weighted short position in the stocks whose previous day's high-to-close price differences were relatively small. The portfolios do not suggest any initial investment, since the total values of the long and the short positions are equal. c) Portfolio based on comparison of low-to-close and high-to-close price differences: Portfolio L-H ("Low-to-close ­ High-to-close price differences"): Portfolio implying an equally-weighted long position for a day's opening session in the stocks whose previous day's high-to-close price differences were greater or equal to their previous day's low-to-close price differences, and an equally-weighted short position for the day's opening session in the rest of the sample stocks. The idea of this portfolio is based on jointly exploiting the 'reversal' versions of Hypotheses 1a and 1b. This portfolio too, does not suggest any initial investment, since the total values of the long and the short positions are equal. According to the definitions, the portfolios in paragraphs (a) and (b) refer to testing Hypotheses 1a and 1b, respectively, while the portfolio in paragraph (c) tests for the joint effect of both tendencies, the issue presented and discussed in Table 2 in the previous Subsection. Table 3 concentrates the daily performance measures over the sampling period for all the five portfolios. Strikingly, all the portfolios yield significantly positive mean daily returns. These results, first of all, provide a strong support for both research hypotheses, even when these are separately tested. That is, opening stock returns are significantly higher following the days with relatively large high-to-close price differences (with respect to the sample's average or median) and also following the days of the relatively small low-to-close price differences. Moreover, from the practical point of view, at least if the trading commissions are not a problem, the five portfolios represent potentially profitable investment strategies. The mean opening returns of about 0.1 percentage point (or even smaller) may, at the first glance, seem not quite impressive, but since we are talking about single-day opening returns, the mean annual return of about 25% on Portfolio L-H or about 30% on portfolio HA, for example, look promising (recall that the portfolios do not request any initial investments and yield significantly positive returns). Table 3. Historical performance measures of the portfolios based on the "overnight reversals" stock price behavior Daily-adjusted Portfolio performance measures (opening returns) over the sampling period (2456 days) portfolios Mean, % Standard Deviation, % t-statistic (Mean=0) Portfolio LA 0.040 0.478 ***4.10 Portfolio LM 0.045 0.406 ***5.50 Portfolio HA 0.104 0.466 ***11.09 Portfolio HM 0.091 0.388 ***11.60 Portfolio L-H 0.090 0.541 ***8.23 Asterisks denote two-tailed p-values: *p<0.10; **p<0.05; ***p<0.01. Overall, the results in this Section strongly indicate that the "end-of-the-day" stock price moves tend to be reversed in the next day's opening session. Investment strategies built upon the expectation of "overnight reversals" may, therefore, possess a non-negligible potential. Conclusion The main goal of the present study is to shed light on the dynamics of stock price reversals and namely, on the overnight reversals. I expect that following some relatively large price moves towards the end of the trading days, there may be price reversals during the next days' opening sessions. To account for price changes at the end of the trading days, I employ high-to-close and low-to-close price differences. I employ intraday price data on thirty stocks currently making up the Dow Jones Industrial Index, and find supporting evidence for my research hypothesis. For each trading day, I compare each stock's high-to-close and low-to-close price changes, and also compare them to the same day's average and median changes for the total sample of stocks, and document that opening returns tend to be higher following the days with relatively large high-to-close price changes (price decreases at the end of the day), and lower following the days with relatively large low-to-close price changes (price increases at the end of the day). These findings imply that stock price changes towards the end of a trading day may contain an element of overreaction to be reversed right at the beginning of the next trading day. Such "overnight reversals" price behavior seems to contradict the market efficiency. Furthermore, I test if on the basis of these findings it is possible to define potentially profitable investment strategies. I construct a number of portfolios based on the opening trading sessions and involving a long position in the stocks on the days when, according to the findings, their opening returns are expected to be high and a short position in the stocks on the days when, according to the findings, their opening returns are expected to be low. All the portfolios are found to yield significantly positive returns, providing an evidence for the practical applicability of the "overnight reversals" pattern in stock prices. Overall, my findings amplify the results documented in the previous literature with respect to the profit potential embedded in the short-term stock price reversals. Lehmann (1990); Jegadeesh (1990), and Conrad et al. (1994) analyze stock price reactions to news and conclude that these are usually too strong (overreaction), and therefore, short-term price reversals may be generally expected. Zawadowski et al. (2006), and Grant et al. (2005) concentrate on especially short time intervals, and document intraday stock price reversals. In this study, I make an effort to "move one step forward" and show that there are consistent (overnight) stock price reversals between consecutive trading days, or in other words, that opening stock returns tend to be higher following the days that were closed with price decreases and lower following the days that were closed with price increases. Journal of Advanced Studies in Finance To summarize, at least in a perfect stock market with no commissions, the daily-adjusted strategies based on the expectations of the "overnight reversals" look promising. This may prove a valuable result for both financial theoreticians in their eternal discussion about stock market efficiency, and practitioners in search of potentially profitable investment strategies. Potential directions for further research may include expending the analysis to other stock exchanges and greater samples, though in the latter case some care has to be taken when defining the comparative benchmarks for high-to-close and low-to-close price difference measures, and also applying similar kind of analysis to longer time intervals.

Journal

Journal of Advanced Studies in Financede Gruyter

Published: Dec 1, 2012

There are no references for this article.