تجزیه و تحلیل فنی روزانه ذخایر فردی در بورس اوراق بهادار توکیو

This paper conducts an intraday technical analysis of individual stocks listed on the Nikkei 225. In addition to the price-based technical rules popularly examined in the literature, we uniquely propose and statistically investigate technical rules that utilize information regarding (1) the order-flow imbalance and (2) the order-book imbalance. Technical analysis using the imbalance-based trading rules is motivated by the evidence presented first in this paper that short-term returns can be predicted from the information regarding the order-flow and order-book imbalances for more than half of Nikkei 225-listed stocks. However, we demonstrate that no strategies, including limit order trading where trading signals are derived from the order-book imbalance, beat the buy-and-hold strategy within our sample. The results imply that past prices and demand/supply imbalances do not contribute to profiting in intraday trading and that non-execution and picking-off risks are too large for limit order trading to be profitable in our sample.

Technical analysis involves the use of historical market data, such as price, volume, and other observables, to predict future returns in financial assets. This technique has been widely used by financial market professionals and much investigated by academic researchers. For example, the studies of survey data conducted by Lui and Mole, 1998, Menkhoff and Taylor, 2007 and Menkhoff, 2010 demonstrate that financial market professionals rely heavily upon technical trading strategies to form their own trading plans.1 Numerous empirical studies of technical analysis have also been carried out, particularly since the 1960s.2 Thus far, real market professionals and academic researchers have paid particular attention to past price data in order to evaluate the profitability of technical trading rules. However, in addition to price and volume data, other market data such as order flows and order-book information, have recently been made available to the public in real time.3 Since order flows and the state of the order book provide information about likely price movements, technical trading strategies that are formulated based upon such information would be expected to yield profits. This paper thus aims to investigate the profitability of the trading strategies based upon the order flows and order-book information as well as past prices by utilizing data from the limit order book and transactions in individual stocks on the Tokyo Stock Exchange. In addition to linking past prices to technical trading strategies, as has been typically carried out by previous studies, the present paper makes a unique contribution to the body of knowledge on this topic by relating (1) the order-flow imbalance and (2) the order-book imbalance to the profitability of intraday technical trading rules. As indicated by Handa and Schwartz (1996, p. 1835), investigating the profitability of trading strategies based upon order-book information provides the rationale not only for limit order trading, which is important to stock traders, but also for the viability of an order-driven market. In addition, technical traders in reality would have more trading options if we were able to provide profitable technical rules based on order-flow information. This paper is also motivated by the several previous research studies that demonstrate that order-flow and order-book imbalances have a predictive ability with respect to short-horizon returns. For example, selected papers, such as those of Chordia et al., 2005, Su et al., 2010 and Visaltanachoti and Yang, 2010, empirically demonstrate that short-horizon returns are predictable from the order-flow imbalance, which is typically defined as the number of seller-initiated trades subtracted from the number of buyers (or the dollars received by sellers subtracted from the dollars paid by buyers). When the order-flow imbalance is positive, meaning that the pressure to buy dominates the market, the price tends to go up, and vice versa. Furthermore, Chordia and Subrahmanyam, 2004 and Su et al., 2010 find that trading profits are generated from the order-flow imbalance. These empirical works confirm the significant relation among the order-flow imbalance, the return predictability, and the trading profitability in the short horizon, and imply that intraday technical trading rules, when combined with the order-flow imbalance, can be profitable. Second, previous studies demonstrate significant relations among the current order imbalance, namely the difference between the current market depths of the buy side and the sell side, investors’ trading decisions, and stock price dynamics. For example, Handa et al. (2003) indicate that stock investors tend to place more aggressive orders, such as market orders, as the order-book depth on the same side of the order becomes thicker compared with that on the other side, and less aggressive orders, such as limit orders outside the spread, as it grows thinner.4 The results of Handa et al. (2003) resemble the findings of several empirical studies such as Biais et al., 1995, Griffiths et al., 2000 and Ranaldo, 2004 that explain as follows. Real stock investors face a trade-off between an advantageous price and the risk of non-execution. When investors submit a market order rather than a limit order, they can certainly execute the order, but they give up obtaining a superior execution price. Meanwhile, investors who place a limit order obtain a favorable execution price, but at a higher risk of non-execution and of failing to earn any profits. Such a risk of non-execution increases as the depth on the same side of the order becomes thicker compared with that on the other side. When an agent recognizes a thicker order book, the agent places more aggressive orders. When the book is thinner, investors tend to assume that more aggressive orders are coming from the opposite side, and thus a limit order is more likely to be executed, yielding more favorable prices for investors at a reduced risk of non-execution as a result of placing less aggressive orders. This trading strategy indicates that when the bid side becomes thicker than the ask side, for example, market buy orders are more likely to be observed, pushing up future prices and making return predictable from the order-book imbalance.5 In addition, Osler (2003) demonstrates that the limit order book is a key source for predicting short-term future returns,6 while Harris and Panchapagesan (2005) show that traders who trade ahead of the heavy side of the book are capable of making a positive profit. Thus, previous studies imply certain profitable opportunities from intraday technical rules in which we derive trading signals from the imbalance of the limit order book. This paper evaluates the intraday trading performances of our technical rules in individual stocks on the Tokyo Stock Exchange. We conduct this evaluation by using an order book and transaction data on individual stocks listed on the Nikkei 225 from September 1, 2006 to August 31, 2007, distributed by Nikkei Media Marketing, Inc., an information vendor in the Nikkei Group. We apply both the Reality Check bootstrap procedure proposed by White (2000) and the Superior Predictive Ability (SPA) test created by Hansen (2005) in order to evaluate the profitability of our trading strategies as well as to reduce the data-snooping problem, which may occur if profitable rules are found by pure chance.7 We first provide evidence that order-flow and order-book imbalances can, in fact, predict short-horizon returns on the Japanese stock market during our sample period. We carry out this procedure before the technical analysis, as little empirical evidence exists in the literature regarding the predictability of the short-horizon returns of individual stocks on the Japanese stock market. However, in contrast to the presented evidence, we demonstrate that none of the technical trading strategies formulated based on past prices, the order-flow imbalance, and information on the demand/supply imbalance in the limit order book generate higher profits compared with a buy-and-hold strategy once the effect of data snooping has been accounted for. At the end of the paper, we confirm the robustness of our results with a sample of different periods that have a higher Nikkei 225 index return volatility, in which extra risk may be added to dynamic strategies, implying a higher risk premium and therefore greater predictability. We emphasize the following two points regarding our results. First, as implied from the results of the predictability of the short-horizon returns, the best technical rules based on order-flow and order-book imbalances as well as past price information actually generate positive trading profits in our sample. However, the profits are too small and not sufficiently larger than those derived from a buy-and-hold strategy. Second, the unprofitable technical rules include limit order trading, in which trading signals are derived from the order-book imbalance.8 Thus, the results of this paper also suggest that non-execution risk and picking-off risk are too large for limit order trading to be profitable during our sample periods.9 Although our most significant contribution is that we propose and statistically investigate unexamined technical rules that are formulated based on order-flow and order-book imbalances, while also taking into account the effect of data snooping, this paper contains four other contributions. First, we demonstrate the predictability of the short-horizon returns of individual stocks from order-flow and order-book imbalances on the Japanese stock market during our sample period. Second, rather than using stock indices as is typical in the literature, we utilize individual stocks in our analyses. Such an approach has the following two advantages. First in reality, certain traders, such as certain day traders, still utilize technical strategies on individual equities. Thus, we provide a way for certain types of investors, who trade individual securities, to make profits. Second, our investigation into individual stocks also provides an approach to better manage the portfolios of investors trading a set of individual stocks, such as those managing ETFs (Exchange Traded Funds), whose trading volumes have recently been expanding rapidly. In terms of our third contribution, we investigate the performance of technical trading rules in terms of firm size. The size-stratified analyses provide insights into whether the differences in the results may be attributed to information asymmetry in the market, institutional herding, or order splitting. For example, Chordia et al., 2005 and Chordia et al., 2008 demonstrate a stronger predictability for short-horizon returns, such as those 5 min in length, for smaller firms in their sample than for larger firms. Generally, larger firms possess a superior information environment, resulting in less information asymmetry in the market. If this situation is indeed the case, the returns of smaller firms are more predictable. However, as Chordia and Subrahmanyam (2004) find, the imbalance has a stronger autocorrelation for large firms than it does for small firms, suggesting a greater influence of order splitting in large firms.10Chordia et al. (2005) suggest that this imbalance also occurs because of the greater influence of institutional herding. Further, in the research of Chordia and Subrahmanyam (2004), the persistence of the imbalance generates a stronger return predictability, which implies that if there is a certain form of herding or order splitting in larger firms, even in intraday trading, the return predictability is stronger for larger firms. Certain previous papers, such as Shynkevich (2012) and Marshall et al. (2008a), conduct size- or industry-stratified technical analyses. Rather than examining the daily time-series of industry or sector portfolios, our paper relates order-flow and order-book imbalances to the technical rules and investigates the intraday performances of individual stocks on the Japanese stock market. Fourth, this paper reports on intraday technical analysis, which has garnered recent attention from academic researchers for two reasons. First, a daily trading horizon is the frequency that is generally examined in previous technical analyses; however, most studies find that technical trading rules are not profitable when transaction costs are included.11 The results suggest that trading only once per day may be too slow to capture profit opportunities. Second, as shown in several studies that survey market participants, such as those of Lui and Mole, 1998 and Menkhoff and Taylor, 2007, and Menkhoff (2010), the shorter the trading horizon, such as the intraday, the greater importance that financial market participants place upon technical trading and the less upon fundamental analysis. Nonetheless, relatively few papers analyze trading profits from technical indicators using ultra high-frequency data in stock markets. Marshall et al. (2008b) use 5-min intervals for a composite price dataset of S&P’s Depository Receipts. They demonstrate that none of their 7846 rules is able to beat the market once the data-snooping problem is taken into account. Our paper differs from that of Marshall et al. (2008b), as we examine a dataset of individual stocks on the Japanese stock market as well as order imbalance technical rules. The remainder of this paper is structured as follows. Section 2 introduces our dataset and summary statistics for returns of our individual stocks. The section also statistically examines the relation between returns and order-flow and order-book imbalances. Section 3 introduces the technical trading rules used in this paper, while Section 4 describes the procedures for White’s Reality Check bootstrap and Hansen’s SPA. Section 5 outlines the empirical tests. Section 6 concludes.

This paper empirically investigates the trading performances of intraday technical rules in individual stocks on the Japanese stock market. In addition to the price-based trading rules commonly examined in previous papers, we make a unique contribution to the body of knowledge on this topic by proposing and statistically investigating the technical rules that are formulated based on (1) the order-flow imbalance and (2) the order-book imbalance. We first provide evidence of the short-term return predictability for more than half of the Nikkei 225-listed stocks in our sample, as little empirical evidence exists in the literature regarding the predictability of the short-horizon returns of individual stocks on the Japanese stock market. We then analyze the profitability of 5081 technical trading strategies for 207 individual stocks listed on the Nikkei 225 from September 1, 2006 to August 31, 2007. White’s Reality Check and Hansen’s Superior Predictive Ability bootstrapping procedures are applied to the 5-min returns series in order to reduce the data-snooping problem. We demonstrate that all technical trading strategies fail to outperform the buy-and-hold strategy, indicating that information on past prices and on demand/supply imbalances in the order book are not related to superior technical trading profitability. Our results imply that transaction costs and the risks of non-execution and picking-off are too large for our technical rules to be profitable during our sample periods. We also demonstrate that our results are robust with sample periods of higher return volatility, covering September 1, 2005 to August 31, 2006. Our results come from a sample of the Japanese stock market over 2 years, which may feature different risk premiums in the market. Investigating the profitability of the order-flow and order-book imbalance technical rules in other years, exchanges, and countries as well as longer samples is an important subject of future research.