Tests of technical trading rules in the Asian-Pacific equity markets: a bootstrap approach.

Lento, Camillo

ABSTRACT

This study examines the effectiveness of nine technical trading rules in eight Asian-Pacific equity markets for periods ranging from January 1987 to November 2005. The annualized returns from each trading rule are compared to a naive buy-and-hold strategy to determine profitability. The TSEC, Straits Times, Hang Seng, Jakarta, KOSPI and the BSE emerge as equity markets where technical trading rules may be profitable. There is no evidence of profitability for the other two markets, the Nikkei and the All Ordinaries. Disregarding statistical significance, the results reveal that 56 out of the 72 (77.8 per cent) trading rule variants tested on all data sets were profitable after accounting for

transaction costs. The results are important because they provide investors with information about the Asian-Pacific equity markets that can be used to determine optimal asset allocations and to further diversify portfolios.

INTRODUCTION

Technical analysis is considered to be one of the earliest forms of investment analysis with its origins dating back to the 1800s. It was one of the first forms of investment analysis mainly because stock prices and volume levels have been publicly available prior to other types of financial information. Technical analysts search the past prices of a time series for recognizable patterns that have the ability to predict future price movements and earn abnormal returns. Various trading rules and indicators have been developed based on each identifiable pattern. The belief that historical data can be used to identify patterns that predict security movements violates the random walk hypothesis and the weak form of market efficiency. According to efficient market theorists, technical analysis will not be able to generate abnormal returns in an efficient market. However, the relatively new and emerging equity markets in the Asian-Pacific region have not been tested extensively to determine whether various types of technical trading rules can be used to earn abnormal returns.

There have been a number of studies conducted on trading rules in the North American equity markets. Alexander (1964) and Fama and Blume (1966) were two of the first to test technical trading rules in the United States. Both of these studies suggest that excess returns could not be realized by making investment decisions based on the movements of certain sizes after adjusting for transaction costs. The number of influential studies that support trading rules grew in the 1990s. Some of these studies include Jegadeesh and Titman (1993), Blume, Easley, and O'Hara (1994), Chan, Jagadeesh, and Lakonishok (1996), Lo and MacKinlay (1997), Grundy and Martin (1998), and Rouwenhorst (1998). Studies that signify the informational content of technical trading rules and patterns include Lo, Mamaysky and Wang (2000), Brock, LeBaron and Lakonishok (1992), GenOay (1999), Lisi and Medio (1997), and Allen and Karjalainen (1999).

There have been a limited number of studies conducted on technical trading rules in the Asian-Pacific markets. Bessembinder and Chan (1995) were the amongst the first to report that moving averages and the trading range break-out rule are useful for forecasting index returns for a group of Asian stock markets. Los (2001) argues that Asian stock markets exhibit strong price trend behaviour and suggests that trading rules with first order Markov filters can be used to profitability exploit trends. Ratner and Leal (1999) tested variable moving average rules in Latin America and Asia and found Taiwan and Thailand as the only two profitable markets.

The purpose of this study is to examine the profitability of various technical trading rules in a number of Asian-Pacific equity markets. Profitability is defined as the ability to earn annualized returns in excess of the buy-and-hold trading strategy. The statistical significance (p-value) of the returns is assessed through a bootstrap simulation. In addition, the robustness of the results is tested through sub-period analysis.

Nine technical trading rules are employed in an attempt to exploit trends in the series of returns from eight Asian-Pacific equity markets. The results demonstrate, on average, that superior profits (after trading costs) can be achieved by technical trading rules over the buy-and-hold trading strategy in certain countries, mainly Bombay, Hong Kong, Indonesia, Korea, Singapore, and Taiwan. Disregarding statistical significance, 56 of the 72 (77.8 per cent) technical trading rules were able earn excess returns, consistent with the findings of Brock, LeBaron and Lakonishok (1992) (referred to as BLL from hereon in) on the Dow Jones Industrial Average and Ratner and Leal (1999) in the emerging markets. Furthermore, 80 per cent of all buy signals were able to correctly predict price movements at the one- and ten-day lag, suggesting that trading rules generate relevant market timing information.

This study differs from the current literature because it provides a more comprehensive test of technical trading rules on the Asian-Pacific equity markets. No other study comprehensively tests as many trading rules on such a large number of equity markets: nine variants of three trading rules (moving average, trading range break-outs and filter rules) on eight Asian-Pacific equity markets. This study also offers more recent data and a different methodology than the two prior studies that focused on the Asian-Pacific markets (Bessembinder and Chan (1995) and Los (2001)). As such, this study contributes to the overall understanding of the efficiency and price behaviour of these markets. An understanding of the Asian-Pacific equity markets is important as they are quickly developing into a very significant portion of the global equity market.

The remainder of the paper is organized as follows. The next section describes the trading rule strategies. Section 3 described the data. Section 4 explains the methodology. Section 5 presents the results. Conclusions and recommendations for future research in Section 6.

TRADING RULES

Trading rules can be grouped into three classes: market structure, market sentiment, and flow of funds. This study tests the most common market structure trading rules. The three rules tested are moving average cross-over rules, filter rules (momentum strategies) and trading range break-out rules. BLL discuss the potential biases that can arise from identifying and testing patterns in security returns in the same dataset. As such, the same trading rules as BLL along with three common filter rules are tested. This will help reduce the possibility of data snooping as the datasets are not searched for successful trading rules ex-post. Testing the trading rules on subsets for robustness also mitigates the effects of data mining on the overall conclusions.

A moving average cross-over (MAC-O) rule compares a short moving average to a long moving average. The MAC-O rule tries to identify a change in a trend. There are two categories of the MAC-O rule: variable length moving average (VMA) and fixed length moving average (FMA). The FMA stresses that the returns for a few days following the crossing of the moving averages should be abnormal. The VMA generates a buy (sell) signal whenever the short moving average is above (below) the long moving average. This study tests the VMA rule based on the following buy and sell signals:

Formula (1)

[[summation].sup.S.sub.s=1] [R.sub.i,t]>/S [[summation].sup.L.sub.l=1] [R.sub.i,t-1]>/L = BUY Formula (1)

[[summation].sup.S.sub.s=1] [R.sub.i,t]>/S [[summation].sup.L.sub.l=1] [R.sub.i,t]>/L = Sell, Formula (2)

where [R.sub.i,t] is the log return given the short period of S (one or five days), and [R.sub.i,t-1] is the log return over the long period L (50, 150 or 200 days). These are the same buy and sell signal used by Ratner and Leal (1999) and various other researchers. The following short, long combinations will be used to test the VMA: (1, 50), (1, 200) and (5, 150).

Filter rules generate buy and sell signals based on the following logic: Buy when the price rises by f per cent above the most recent trough and sell when the price falls f per cent below its most recent peak. The filter size (Y) is the parameter that defines a filter rule. This study tests the filter rule based on three parameters: one-per cent, two-per cent, and five-per cent.

The trading range break-out (TRB-O) rule, also referred to as resistance and support levels, generates a buy signal when the price breaks-out above the resistance level and a sell signal when the price breaks below the support level. The resistance level is defined as the local maximum, and the support level is defined as the local minimum (BLL). At the resistance (support) level, intuition would suggest that many investors are willing to sell (buy). The selling (buying) pressure will create resistance (support) against the price rising (falling) above the peak (trough) level. The TRB-O rule is examined by calculating the local maximum and minimum based on 50, 150 and 200 days as defined in Formula 3.

[Pos.sub.t+1] = Buy, if [P.sub.t] > Max {[P.sub.t-1], [P.sub.t-2], ..., [P.sub.t-n]} [Pos.sub.t+1] = [Pos.sub.t], if [P.sub.t] > Min{[P.sub.t-1], [P.sub.t-2], ..., [P.sub.t-n]} [P.sub.t] Max {[P.sub.t-1], [P.sub.t-2], ..., [P.sub.t-n]} [Pos.sub.t+1] = Sell, if [P.sub.t] < Min{[P.sub.t-1], [P.sub.t-2], ..., [P.sub.t-n]} Formula (3)

where [P.sub.t] is the stock price at time t.

DATA DESCRIPTION

The technical trading rules are tested on eight Asian-Pacific equity markets, including the more highly developed Nikkei and Hang Seng markets. The data sets of the eight equity markets tested are described in Table 1.

The trading rules can be calculated at various data frequencies. Investors can use high-frequency data, such as intra-day, or longer horizons, such as weekly or yearly, when using the trading rules. The frequency selected by a technical investor depends on many different factors and personal preferences. This research study utilizes daily closing prices for the stock market indices over a minimum of seven years. A seven-year period provides a sufficient number of daily observations to allow for the formation, recurrence and investigation of the technical trading rules. The daily returns are calculated as the holding period return of each day as follows:

[r.sub.t] = log ([p.sub.t]) - log ([p.sub.t-1]) Formula (4)

where [p.sub.t] denotes the market price.

METHODOLOGY

Trading Rule Profitability

The profitability of the trading rules is determined by comparing the returns generated by the trading signals to the buy-and-hold strategy returns. The methodology relies on this relatively simple technique for analyzing the profitability of the trading rules because of the possible problems related to non-linear models such as computational expensiveness, overfitting, data snooping and difficulties interpreting the results. See White (2005) for a thorough discussion of these issues. As such, the returns are subject to sophisticated tests of significance. The returns from the buy-and-hold strategy are calculated by investing in the security at the beginning of the data set, given the trading rule parameters, and holding the security until the end of the data set. For example, no trading signal can be generated until the 50th day with a 1-day, 50-day MAC-O rule. Therefore, the buy-and-hold returns will be calculated commencing the 50th trading day.

The trading rule returns are also calculated in a relatively simple manner. The returns resulting from the MAC-O rules are based on the variable moving average signals. More precisely, when a buy signal is triggered as per Formula 1, the investor will take a long position, and returns will be calculated at the market rate. When a sell signal is triggered as per Formula 2, the investor is out of the market and returns will be based on a notional interest rate (3 per cent APR or 0.0089 daily EAR). A nominal interest rate is used because the data sets are not adjusted for inflation.

The returns resulting from the filter rule and TBR-O rule are calculated in a slightly different manner. At the beginning of the trading period, the investor will be short and earn the notional interest rate. To minimize the measurement error due to non-synchronous trading made evident by Scholes and Williams (1977) the investor will be long the market one day after the trading signal is generated. Therefore, once a buy signal is generated, the investor will be long on the following day, and returns will be calculated based on the market returns. Finally, if the investor is long (short), and a buy (sell) signal is generated, the position is carried forward.

Similar to Gencay (1998b), the returns generated from the trading rules are adjusted for transaction costs. Both the bid-ask spread and brokerage trading costs are included into the total transaction cost. The bid-ask spread for an exchange traded fund of the index is used as a proxy for the actual index. The return will be adjusted downward by 0.99859 when a trade is triggered. This adjustment factor approximates the average transaction costs for these securities. See Ratner and Leal (1999) for a summary of transaction costs in Asian equity markets.

The significance of the results is tested by using the bootstrap approach developed by Levich and Thomas (1993). This approach, first, observes the data set of closing prices, with the sample size denoted by N+1 that corresponds to a set of N returns. The [m.sup.th] (m=1, ... ,M) permutation of these N returns (M=N!) is related to a unique profit measure (X[m, r]) for the [r.sup.th] trading rule variant (r=1, ... ,R) used in this study. Thus, for each variable, a new series can be generated by randomly reshuffling the returns of the original series.

From the sequence of M returns, the starting and ending points of the randomly generated time series are fixed at their original values. This maintains the distributional properties of the original data. However, the time series properties are random. In this bootstrapping simulation one can thus generate one of the various notional paths that the returns could have taken from time t (starting day) to time t+n (ending day). The notional paths are generated fifty times for each data set. Technical trading rules are then applied to each of the fifty random series and the profits X[m, r] are measured. This process generates an empirical distribution of the profits. The profits calculated on the original data sets are then compared to the profits from the randomly generated data sets. A simulated p-value is produced by computing the proportion of returns generated from the simulated series that is greater than the return computed with the actual series.

The null and alternative hypotheses are given by:

[H.sub.0]: the trading rules provide no useful information.

[H.sub.1]: the trading rules provide useful information.

Robustness testing will be performed to mitigate the effects of data mining and to further analyze the significance of the trading rule profits. To test the returns for robustness, returns will be calculated on three sub-periods of the original data. The sub-periods are determined by arbitrarily dividing the data sets into thirds and then testing for structural breaks between the subsets. The Chow Test is used to test for structural breaks. The subsets will be used to test for robustness if the parameters of each subset are determined to be non-stationary. Three new subsets are selected and re-tested if the parameters of the subsets are constant.

The returns from each trading rule and the buy-and-hold strategy, along with the Sharpe ratio, are computed for each sub-period. Consistent excess returns and stable Sharpe ratios across the sub-periods are associated with robust returns.

Sign Prediction Ability of Trading Rules

The effectiveness of the trading signal's ability to predict future price movements in the equity markets given a one-day and ten-day lag is also tested. A one-day and ten-day lag is utilized because one lag provides a measure of immediate effectiveness (one-day), while the other is a more flexible measure (ten-day). This is similar to the BLL evaluation methodology.

Assuming a trading signal is generated at time t, the one-day lag is defined as the log return at time t+1. The ten-day lag is defined as the holding period return for the ten days immediately following time t. As such, a buy signal is correct if the one or ten day holding period return following the signal is positive. A sell signal is correct if the one or ten day holding period return following the signal is negative. For example, the one-day lag return is calculated as follows: if a trading rule generates a buy signal for a security on June 1st, the signal will be deemed to be correct if the price increases in the following day, June 2nd. Conversely, a sell signal is correct if the price of the security decreases the day following a trade signal. Based on this rational, a predictive value (PV from hereon in) can be calculated. The PV is calculated as follows:

([CS.sub.t]) / ([CS.sub.t] + [IS.sub.t]) Formula (5)

where [CS.sub.t] denotes the number of correct buy signals give the time lag (t) and [IS.sub.t] denotes the number of incorrect buy signals given the time lag (t). The PV measure is used extensively in laboratory tests for medical research studies; however, it is an applicable measure for the purposes of this study as well.

Along with the predictive values for each buy and sell signal, the binomial probability distribution (BPD) will be used to calculate the probability of the PV occurring by chance. The BPD probability will be presented so that it can be interpreted similar to a p-value. The BPD is used for events with dichotomous results, where the probability of success is constant for each trial, and trials are independent. The signals tested meet these requirements.

The effectiveness of the trading rule sign prediction ability will also be tested by analyzing the aggregate average daily returns that follow a signal. Naturally, it is expected that a large and positive daily return will follow a buy signal, and a small or negative return will follow a sell signal. The significance of the returns will be determined by testing for a significant difference between returns following the buy and sell (buy-sell) signals. The trading rules can forecast future movements of security returns if the difference between the buy-sell returns is positive and significant. The t-stat will be calculated as follows:

[[mu].sub.b] - [[mu].sub.s]/[square root of ([[sigma].sup.2.sub.b]/[[eta].sub.b]) + [[sigma].sup.2.sub.S]/[[eta].sub.s]) Formula (6)

Refer to BLL for a detailed discussion of the t-test methodology. Note that this test statistic does not always conform to the student distribution. However, an approximation for the degrees of freedom was developed by Satterthwhaite (1946). If the number of observations is sufficiently large, this test statistic will converge to a standard normal distribution and the t-table critical values can be used.

EMPIRICAL RESULTS

Profitability of Trading Rules

The profitability of the technical trading rules is presented in Table 2. The resulting p-values from the bootstrapping simulation are also presented in Table 2. If the original return has a rank of 100, then the return was the highest of any of the randomly generated returns, and has a corresponding p-value of 0.00. A rank of fifty reveals that half of the randomly generated returns were greater than the original return, resulting in a p-value of 0.50.

The technical trading rules performed best on the Strait Times and TSEC markets as all nine variants generated returns in excess of the naive buy-and-hold trading strategy. The trading rule performed the worst on the All Ordinaries data set as only two of the nine trading rules generated excess returns. Overall, 56 of the 72 (77.8 per cent) trading rule variants tested on all data sets were able to earn excess returns. The bootstrapping simulations reveal only 27 of the 56 (47.4 per cent) excess returns are statistically significant at the five per cent level of significance

In general, the MAC-O trading rules performed the best of the three rules as 22 of the 24 (91.7 per cent) tests generated excess returns, twelve of which were significant at five per cent. More specifically, the MAC-O (1, 50) trading rule performed the best as all nine variants tested outperformed the buy-and-hold trading strategy in all nine Asian-Pacific markets. The MAC-O (1, 50) earned excess returns in the range of 1.8 to 32.6 per cent per annum. The MAC-O (1, 200) trading rule also earned excess of returns ranging from 1.5 to 10.9 per cent per annum for eight of the nine markets; excess returns were not available only on the All Ordinaries market index.

The filter rules earned excess returns for 17 of the 24 (70.8 per cent) variants tested, however only 11 of the 24 filter rules were significant at five per cent level of significance. None of the filter rules (one, two, or five per cent) were able to generate excess returns on the Nikkei, and only the one-per cent filter rule was able to beat the buy-and-hold trading strategy on the All Ordinaries. However, the filter rules performed well on the remaining market indices as 16 of the 18 earned excess returns.

The TRB-O rules also beat the market in 17 of the 24 (70.8 per cent) variants tested, but only four of the TRB-O rules were significant. The TRB-O (50 days) was able to beat the market in 7 of the 8 variants tested. None of the TRB-O rules were able to earn excess returns on the All Ordinaries, and only the TRB-O (50 days) was able to beat the market on the BSE. However, the TRB-O generated excess returns in 16 of the 18 variants tested.

In general, the results suggest that trading rules based on short-term momentum are better at generating statistically significant excess returns. Excluding the Nikkei, both the MAC-O (1, 50) and the one per cent filter rule consistently provide statistically significant excess returns. The 50-day TRB-O rule is also the most profitable of all three TRB-O variants. The bootstrapping simulation provides some support against the weak form of the EMH revealing that the MAC-O (1, 50) and the one per cent filter rule consistently generate significant excess returns.

Similar to GenOay (1998a), the trading rules were tested for robustness on sub-periods. Table 3 present the returns for the sub-periods, along with the Sharpe Ratio for each period.

The sub-period analysis suggests that the returns from technical trading rules are not robust. Overall, 11 of the 72 (15.3 per cent) trading rules tested have positive returns in all three sub-periods. Furthermore, the Sharpe Ratio is not stable and frequently changes across sub-periods that exhibit consistent excess returns. The most robust returns were generated from the MAC-O (1, 50) as three of the eight returns were robust. However, inconsistent return/risk ratios across sub-periods are in line with prior studies. Dooley and Shafer (1983) suggest that the inconsistent return/risk ratios across sub-periods suggest that the returns earned by the profitable technical trading rules over the entire period are risky.

Sign Prediction Ability of Trading Rules

Aside from profitability, this study also seeks to determine the sign prediction ability of the technical trading rules. Sign prediction ability refers to whether the trading rules generate correct buy or sell signals. The PV of each trading signal given a one- and ten-day lag is presented in Table 4. Note that, Table 2 and Table 3 present the results of the VMA from a profitability standpoint, while the sign prediction ability tests are of the FMA.

In total, 288 trading signals were investigated: [9 trading rules x 2 (buy and sell signals) x 2 (lag-one and lag-ten) x 8 data sets]. Overall, 183 of 288 (63.5 per cent) rules yielded a PV greater than 50 per cent. The buy signals were correct more often than the sell signals as 110 of the 144 (76.3 per cent) PVs of the buy signals were greater than 50 per cent, while 73 of the 144 (50.7 per cent) PVs for the sell signals were correct greater than 50 per cent. However, 50 per cent is not a high benchmark for a PV. The precision of each signal can be paralleled to tossing a coin. Like a coin, the signal can either be correct or incorrect. Therefore, 50 per cent can be easily obtained by chance. The BPD was used to assess the probabilities of each correct percentage occurring by chance. Table 3 presents the BPD probability of the result that can be interpreted in a similar fashion a p-value. Based on a five per cent level of significance, 68 of the 288 (23.6 per cent) signals provided relevant information regarding future price movements.

The aggregate daily returns that follow the buy and sell signals and the Buy-Sell t-stat are presented in Table 5 (one-day lag in Panel A and ten-day lag in Panel B). The daily returns following the signals should provide the same conclusion regarding the informational content of trading rule signals as the PV analysis.

The returns following a buy signal were positive in 58 of 72 (80.0 per cent) of cases and the returns after a sell signal were negative in 50 of the 72 (69.4 per cent) cases given a one-day lag. At the ten-day lag, the returns following a buy signal were positive in 58 of 72 (80.0 per cent) of cases and the returns after a sell signal were negative in 47 of the 72 (65.2 per cent) cases. The results are negatively affected by the Nikkei as 6 of the 18 (33.3 per cent) returns following a buy signal were positive and 7 of the 18 (38.9 per cent) returns following a sell signal were negative. The fact that the trading rules performed the worst on the Nikkei and Hang Seng was expected as they are likely the most efficient and developed Asian-Pacific market tested in this study.

The daily returns can also be compared to the results of the BLL study on the Dow Jones Industrial Average. Table 5 reveals that positive returns follow buy signals at both the one- or ten-day lag on a consistent basis (80 per cent). However, negative returns do not follow a sell signal as consistently as the buy signal at either the one- or ten-day lag. These results are similar to what was found in the U.S. markets by BLL. It appears that more relevant trading information is generated from the buy signals as opposed to the sell signals.

The daily returns corroborate the evidence presented in Table 2 and Table 3. Technical trading rules can provide information that is relevant for timing entry and exit points in certain Asia-Pacific equity markets (sign prediction ability), thus potentially leading to abnormal returns that are in excess of what would be realized through the na<ve buy-and-hold trading strategy. In general, these results are similar to what was discovered by Ratner and Leal (1999) and Bessembinder and Chan (1995).

CONCLUSIONS AND DISCUSSION

An empirical study was conducted to determine if technical trading rules are profitable in the Asian-Pacific equity markets. Profitability was defined as returns in excess of the buy-and-hold trading strategy after accounting for transaction costs. Nine technical trading rules were tested on eight Asian-Pacific equity markets. The results demonstrate, on average, that profits (after estimated trading costs) can be earned by technical trading rules in certain countries, mainly Bombay, Hong Kong, Indonesia, Korea, Singapore, and Taiwan. The results also suggest that buy signals can provide relevant trading information. Based on these results, and similar findings by Bessembinder and Chan (1995) and Ratner and Leal (1999), Bessembinder and Chan (1998) suggest that even if an investor cannot earn a profit after adjusting for transaction costs, a Bayesian investor could alter his asset allocation in response to this information. Therefore, the results of this study may have significant economic implications.

This study differs from the current literature because it provides a more comprehensive test of technical trading rules on the Asian-Pacific equity markets with more recent data and a different methodology. As such, this study contributes to the overall understanding of the efficiency and price behaviour of the Asian-Pacific equity markets. The results of this study are consistent with the reasoning that some of the Asian-Pacific equity markets were informationally inefficient, at least over the period analyzed, as the trading rules were able to earn profits and generate relevant trading information. However, like Bessembinder and Chan (1995), an alternative explanation maybe that the results are sensitive to the round trip transaction cost.

Further research should be conducted to explore the relationship between technical trading rules and market microstructure and order flows. Microstructure can possibly be used as a tool to explain the profitability and predictability of trading rules and market movements. The trading signals generated from the Asian-Pacific markets can also be further processed by applying a combined signal approach (Lento and Gradojevic 2006). Future studies can also explore the investment behaviour of different cultures (i.e. North American, European, and Asian) and the returns to technical trading rules in each respective equity market (i.e. if Asian investors believe more in technical analysis than Europeans do, returns to technical trading rules may be greater in Asian equity markets).

REFERENCES

Allen, F. & R. Karjalainen (1999). Using Genetic Algorithms to Find Technical Trading Rules. Journal of Financial Economics, 51, 245-271.

Alexander, S. (1964). Price Movements in Speculative Markets: Trends or Random Walks, No. 2 in P. Cootnered (ed.), The Random Character of Stock Market Prices. Cambridge, MA: MIT Press.

Bessembinder, H. & K. Chan (1995). The profitability of technical trading rules in the Asian stock markets. Pacific-Basin Finance Journal, 3, 257-284.

Bessembinder, H. & K. Chan, (1998). Market efficiency and the returns to technical analysis. Financial Management, 27 (2), 5-17.

Brock, W., J. Lakonishok, & B. LeBaron (1992). Simple Technical Trading Rules and the Stochastic Properties of Stock Returns. Journal of Finance, 47, 1931-1764.

Chan, L., N. Jegadeesh, & J. Lakonishok (1996). Momentum Strategies. Journal of Finance, 51, 1681-1713.

Dooley, M. P. & J.R. Shaler (1983). Analysis of Short-Run Exchange Rate Behaviour: March 1973 to November 1981 in Exchange Rate and Trade Instability: Causes, Consequences, and Remedies. Cambridge, MA.

Fama, E. & M. Blume (1966). Filter Tests and Stock Market Trading. Journal of Business, 39, 226-241.

Gencay, R. (1998a). The Predictability of Security Returns with Simple Technical Trading Rules. Journal of Empirical Finance, 5 (1998) 347-359

Gencay, R. (1998b). Optimization of Technical Trading Strategies and the Profitability in Security Markets. Economics Letters, 59, 249-254.

Gencay, R. (1999). Linear, non-linear and essential foreign exchange rate prediction with simple technical trading rules. Journal of International Economics, 47, 91-107.

Grundy, B. & S. Martin (1998). Understanding the Nature of the Risks and the Source of the Rewards to Momentum Investing Working paper, Wharton School, University of Pennsylvania.

Jegadeesh, N. & S. Titman (1993). Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency. Journal of Finance, 65-91.

Lento, C. & N. Gradojevic (2006). The Profitability of Technical Trading Rules: A Combined Signal Approach. Journal of Applied Business Research, (in press).

Levich, R. & L. Thomas (1993). The Significance of Technical Trading Rules Profits in the Foreign Exchange Market: A Bootstrap Approach. Journal of International Money and Finance, 12, 451-474.

Lisi, F. & A. Medio (1997). Is a random walk the best exchange rate predictor? International Journal of Forecasting, 13, 255-267.

Lo, A. W., & A.C. MacKinlay (1997). Maximizing Predictability in the Stock and Bond Markets. Macroeconomic Dynamics, 1, 102-134.

Lo, A. W., H. Mamaysky, & J. Wang (2000). Foundations of Technical Analysis: Computational Algorithms, Statistical Inference, and Empirical Implementation. The Journal of Finance 55 (4), 1705-1765.

Los, C. (2001). Nonparametric Efficiency Testing of Asian Stock Markets Using Weekly Data. Centre of Research in Financial Services Murphy, J. (2000). Charting Made Easy. Marketplace Books, 2000

Pesaran, M. H. & A. Timmermann (1992). A Simple Nonparametric Test of Predictive Performance. Journal of Business & Economic Statistics, 10, 461-465.

Ratner, M., & R.P.C. Leal, (1999). Tests of technical trading strategies in the emerging equity markets of Latin America and Asia. Journal of Banking and Finance, 23, 1887-1905.

Rouwenhorst, G. (1998). International Momentum Strategies. Journal of Finance, 53, 267-284.

Scholes, M., & J. Williams (1977). Estimating betas from non-synchronous data. Journal of Financial Economics, 5, 309-327.

Satterthwhaite, R.E. (1946), Biom. Bull. 2:110.

White, H. (2005). Approximate Nonlinear Forecasting Methods. G. Elliot, C.W.J. Gragner, and A. Timmerman, eds., Handbook of Economics Forecasting, (in press).

Camillo Lento, Lakehead University

Table 1--Data Set Descriptions

Country Index Name n Currency

Australia All Ordinaries 5395 Australian
 Dollar

India Bombay Stock 2083 Indian Ruppe
 Exchange (BSE)

Indonesia Jakarta 2030 Indonesian
 Rupiah

Korea Korea Composite 2067 Korean Won
 Stock Price Index
 (KOSPI)

Japan Nikkei 2846 Japanese Yen

Hong Kong Hang Seng 2196 Hong Kong
 Dollar

Singapore Straits Times 4479 Singapore
 Dollar

Taiwan Taiwan Stock 2067 Taiwan
 Exchange (TSEC) Dollar

Country Skewness Kurtosis Period Tested

Australia 0.3643 -0.7093 01/08/87 - 01/11/05

India 1.1715 0.8579 01/01/97 - 01/11/05

Indonesia 1.1862 0.4010 01/01/97 - 01/11/05

Korea 0.1090 -0.1778 01/01/97 - 01/11/05

Japan 0.1202 -1.1654 01/01/95 - 01/11/05

Hong Kong 0.2264 -0.6968 01/01/95 - 01/11/05

Singapore -0.1558 -1.0480 01/12/87 - 01/11/05

Taiwan 0.4970 -0.5101 01/01/97 - 01/11/05

Table 2 - Profitability of Technical Trading Rules (Panel A)

 MA Cross-Over Rule
 Short (days) / Long (days)

Market Index 1 / 50 1 / 200 5 / 150

All Ordinaries (N = 5395)
 Annual Return 13.2 8.1 9.7
 Buy & Hold Return 11.3 10.5 11.1
 Over / (Under) Performance 1.8 (2.3) (1.4)
 No. of Trades 291 120 78
 p-value 0.00 * 0.21 0.24
BSE (N = 2083)
 Annual Return 22.0 12.4 17.1
 Buy & Hold Return 9.8 11 14.5
 Over / (Under) Performance 12.2 1.5 2.6
 No. of Trades 87 48 23
 p-value 0.01 * 0.17 0.29
Jakarta (N =2030)
 Annual Return 38.2 20.6 11
 Buy & Hold Return 5.7 11.3 8.8
 Over / (Under) Performance 32.6 9.3 2.2
 No. of Trades 77 31 28
 p-value 0.03 * 0.12 0.32
KOSPI (N = 2067)
 Annual Return 28.7 25.0 33.7
 Buy & Hold Return 1.2 14.1 10.2
 Over / (Under) Performance 27.5 10.9 23.5
 No. of Trades 97 35 19
 p-value 0.00 * 0.06 0.01 *

 Filter Rule
 (in %)

Market Index 1% 2% 5%

All Ordinaries (N = 5395)
 Annual Return 13.2 9.9 6
 Buy & Hold Return 11.4 11.4 11.4
 Over / (Under) Performance 1.8 (1.5) (5.4)
 No. of Trades 1116 494 57
 p-value 0.00 * 0.06 0.54
BSE (N = 2083)
 Annual Return 13.0 9.4 3.6
 Buy & Hold Return 8.0 8 8.0
 Over / (Under) Performance 5.0 1.4 (4.4)
 No. of Trades 659 442 123
 p-value 0.01 * 0.11 0.54
Jakarta (N =2030)
 Annual Return 59.2 10.2 24.2
 Buy & Hold Return 0.6 0.6 0.6
 Over / (Under) Performance 58.6 9.6 23.6
 No. of Trades 585 395 153
 p-value 0.00 * 0.03 * 0.01 *
KOSPI (N = 2067)
 Annual Return 6.9 (21.0) 6.5
 Buy & Hold Return (0.8) (0.8) (0.8)
 Over / (Under) Performance 7.7 -20.2 7.3
 No. of Trades 722 560 223
 p-value 0.02 * 0.91 0.18

 Trading Range Break-Out
 (days of local max/min)

Market Index 50 150 200

All Ordinaries (N = 5395)
 Annual Return 10.4 6.4 4.7
 Buy & Hold Return 11.3 11.1 10.5
 Over / (Under) Performance (0.9) (4.8) (5.8)
 No. of Trades 454 303 257
 p-value 0.16 0.75 0.89
BSE (N = 2083)
 Annual Return 20.4 7.7 21.7
 Buy & Hold Return 9.8 14.5 11.1
 Over / (Under) Performance 10.6 (6.9) 10.5
 No. of Trades 169 96 82
 p-value 0.06 0.63 0.02 *
Jakarta (N =2030)
 Annual Return 13.4 15.3 8.3
 Buy & Hold Return 5.7 8.8 10.9
 Over / (Under) Performance 7.7 6.5 (2.5)
 No. of Trades 152 8.3 72
 p-value 0.18 0.11 0.50
KOSPI (N = 2067)
 Annual Return 17.6 7.6 6.3
 Buy & Hold Return 1.2 10.2 14.3
 Over / (Under) Performance 16.4 (2.6) (8.1)
 No. of Trades 189 110 97
 p-value 0.07 0.51 0.68

* Significant p-values at the 5% level.

Note that Table 2 presents the number of trades as opposed to
signals. The number of trades is more relevant because transaction
costs are a function of trades, not signals. As noted in the
Section 4, the number of signals does not represent the number of
trades because if an investor is long (short) in the market, no
trade is triggered if a long (short) signal is generated.

Table 2 - Profitability of Technical Trading Rules (Panel B)

 MA Cross-Over Rule
 Short (days) / Long (days)

Market Index 1 / 50 1 / 200 5 / 150

Nikkei (N = 2846)
 Annual Return (1.1) 4.6 8.8
 Buy & Hold Return (4.8) (6.1) (5.0)
 Over / (Under) Performance 3.8 10.7 13.8
 No. of Trades 199 54 27
 p-value 0.31 0.12 0.04 *
Hang Seng (N = 2792)
 Annual Return 10.0 10.5 7.3
 Buy & Hold Return 2.5 0.7 1.7
 Over / (Under) Performance 7.5 9.8 5.5
 No. of Trades 162 46 42
 p-value 0.00 * 0.04 * 0.12
Straits Times (N =4479)
 Annual Return 12.4 8.9 11.5
 Buy & Hold Return 4.6 3.9 3.2
 Over / (Under) Performance 7.8 5.0 8.3
 No. of Trades 268 112 58
 p-value 0.00 * 0.01 * 0.00 *
TSEC (N =2067)
 Annual Return 14.4 0.3 (3.5)
 Buy & Hold Return (11.6) (10.5) (11.5)
 Over / (Under) Performance 26.0 10.8 8.1
 No. of Trades 85 63 38
 p-value 0.01 * 0.14 0.29

 Filter Rule
 (in %)

Market Index 1% 2% 5%

Nikkei (N = 2846)
 Annual Return (16.1) (9.0) (10.4)
 Buy & Hold Return (7.0) (7.0) (7.0)
 Over / (Under) Performance (9.2) (2.0) (3.4)
 No. of Trades 894 543 107
 p-value 0.71 0.60 0.95
Hang Seng (N = 2792)
 Annual Return 9.5 (1.3) 7.6
 Buy & Hold Return 3.3 3.3 3.3
 Over / (Under) Performance 6.1 (4.7) 4.2
 No. of Trades 809 520 145
 p-value 0.00 * 0.43 0.23
Straits Times (N =4479)
 Annual Return 17.1 8.9 7.2
 Buy & Hold Return 5.6 5.6 5.6
 Over / (Under) Performance 11.5 3.3 1.6
 No. of Trades 1068 621 143
 p-value 0.00 * 0.001 * 0.07
TSEC (N =2067)
 Annual Return (0.3) (3.9) 3.3
 Buy & Hold Return (11.5) (4.5) (11.5)
 Over / (Under) Performance 11.3 7.7 14.9
 No. of Trades 678 464 132
 p-value 0.00 * 0.07 0.05 *

 Trading Range Break-Out
 (days of local max/min)

Market Index 50 150 200

Nikkei (N = 2846)
 Annual Return (2.0) 5.3 2.9
 Buy & Hold Return (4.8) (5.0) (6.1)
 Over / (Under) Performance 2.0 10.3 9.0
 No. of Trades 247 140 127
 p-value 0.62 0.12 0.17
Hang Seng (N = 2792)
 Annual Return 7.8 6.2 10.1
 Buy & Hold Return 2.5 1.7 0.7
 Over / (Under) Performance 5.3 4.5 9.4
 No. of Trades 237 126 112
 p-value 0.12 0.16 0.05 *
Straits Times (N =4479)
 Annual Return 14.1 8.5 4.7
 Buy & Hold Return 4.6 3.2 4.0
 Over / (Under) Performance 9.5 5.3 0.7
 No. of Trades 312 182 163
 p-value 0.00 * 0.04 * 0.26
TSEC (N =2067)
 Annual Return 3.2 2.7 (8.4)
 Buy & Hold Return (11.6) (11.5) (10.5)
 Over / (Under) Performance 14.9 8.9 2.1
 No. of Trades 179 90 73
 p-value 0.08 0.27 0.63

* Significant p-values at the 5% level. Note that Table 2 presents
the number of trades as opposed to signals. The number of trades is
more relevant because transaction costs are a function of trades, not
signals. As noted in the Section 4, the number of signals does not
represent the number of trades because if an investor is long (short)
in the market, no trade is triggered if a long (short) signal is
generated.

Table 3 - Profitability of Technical Trading Rules on Sub-Periods
(Panel A)

 Excess Sharpe Ratio
 Return

 All Ordinaries

 03/08/1984 - 27/8/1991
Trading Rule
 MA (1, 50) 12.7 0.0911
 MA (1, 200) 1.8 0.0293
 MA (1, 150) 5.1 0.0411
 Filter Rule (1%) 12.9 0.0909
 Filter Rule (2%) 3.2 0.0625
 Filter Rule (5%) (2.1) 0.0446
 TRB-O (50 days) 7.4 0.0730
 TRB-O (150 days) (0.6) 0.0288
 TRB-O (200 days) (6.3) 0.0102

 Jakarta

 01/07/1997 - 21/3/2000
Trading Rule
 MA (1, 50) 63.1 0.0654
 MA (1, 200) 2.1 0.0046
 MA (1, 150) (8.2) -0.0256
 Filter Rule (1%) 133.0 * 0.1177
 Filter Rule (2%) 31.6 0.0130
 Filter Rule (5%) 33.5 0.0145
 TRB-O (50 days) 13.1 0.0015
 TRB-O (150 days) 17.6 0.0193
 TRB-O (200 days) 6.8 0.0112

 BSE

 01/07/1997 - 2/5/2000
Trading Rule
 MA (1, 50) 11.2 * 0.0185
 MA (1, 200) 3.3 0.0037
 MA (1, 150) (4.2) 0.0128
 Filter Rule (1%) 5.3 -0.0111
 Filter Rule (2%) (2.7) -0.0206
 Filter Rule (5%) (3.0) -0.0191
 TRB-O (50 days) 22.7 * 0.0395
 TRB-O (150 days) (19.7) -0.0139
 TRB-O (200 days) 8.4 0.0147

 KOSPI

 01/07/1997 - 18/4/2000
Trading Rule
 MA (1, 50) 63.9 0.0559
 MA (1, 200) 26.3 0.0596
 MA (1, 150) 37.9 0.0577
 Filter Rule (1%) 27.6 0.0133
 Filter Rule (2%) (26.9) -0.0781
 Filter Rule (5%) 40.1 0.0279
 TRB-O (50 days) 45.7 0.0404
 TRB-O (150 days) 9.7 0.0277
 TRB-O (200 days) (26.4) 0.0083

 Excess Sharpe
 Return Ratio

 All Ordinaries

 28/08/1991 - 28/10/1998
Trading Rule
 MA (1, 50) (2.0) 0.0362
 MA (1, 200) (3.3) 0.0281
 MA (1, 150) (4.7) 0.0223
 Filter Rule (1%) (2.7) 0.0329
 Filter Rule (2%) (3.8) 0.0262
 Filter Rule (5%) (4.8) 0.0207
 TRB-O (50 days) (5.2) 0.0203
 TRB-O (150 days) (7.5) 0.0091
 TRB-O (200 days) (6.4) 0.0126

 Jakarta

 22/03/2000 - 18/02/2003
Trading Rule
 MA (1, 50) 33.2 0.0374
 MA (1, 200) 24.8 0.0163
 MA (1, 150) 21.5 0.0043
 Filter Rule (1%) 29.1 * 0.0226
 Filter Rule (2%) 12.1 -0.0259
 Filter Rule (5%) 36.8 0.0415
 TRB-O (50 days) 26.7 0.0185
 TRB-O (150 days) 11.7 -0.0271
 TRB-O (200 days) (1.8) -0.0646

 BSE

 03/05/2000 - 04/03/2003
Trading Rule
 MA (1, 50) 13.3 * -0.0135
 MA (1, 200) 3.4 -0.0712
 MA (1, 150) 11.5 -0.0246
 Filter Rule (1%) 14.5 -0.0072
 Filter Rule (2%) 15.6 -0.0037
 Filter Rule (5%) 21.8 -0.0137
 TRB-O (50 days) 5.3 * -0.0375
 TRB-O (150 days) 10.0 -0.0302
 TRB-O (200 days) 18.6 0.0116

 KOSPI

 19/04/2000 - 20/02/2003
Trading Rule
 MA (1, 50) 25.3 0.0124
 MA (1, 200) 21.0 0.0041
 MA (1, 150) 33.2 0.0337
 Filter Rule (1%) 11.5 -0.0142
 Filter Rule (2%) 1.8 -0.0335
 Filter Rule (5%) (2.2) -0.0427
 TRB-O (50 days) 5.6 -0.0299
 TRB-O (150 days) 5.5 -0.0297
 TRB-O (200 days) 17.2 -0.0049

 Excess Sharpe Ratio
 Return

 All Ordinaries

 29/10/1998 - 30/11/2005
Trading Rule
 MA (1, 50) (3.8) 0.0399
 MA (1, 200) (5.0) 0.0294
 MA (1, 150) (3.7) 0.0358
 Filter Rule (1%) (9.1) 0.0137
 Filter Rule (2%) (3.8) 0.0397
 Filter Rule (5%) (3.7) 0.0405
 TRB-O (50 days) (4.1) 0.0383
 TRB-O (150 days) (5.6) 0.0259
 TRB-O (200 days) (4.6) 0.0294

 Jakarta

 19/02/2003 - 30/11/2005
Trading Rule
 MA (1, 50) (11.6) 0.1265
 MA (1, 200) (11.1) 0.1049
 MA (1, 150) (18.4) 0.0975
 Filter Rule (1%) 8.8 * 0.1705
 Filter Rule (2%) (32.7) 0.0802
 Filter Rule (5%) (17.3) 0.1061
 TRB-O (50 days) (32.0) 0.0845
 TRB-O (150 days) (15.7) 0.0976
 TRB-O (200 days) (14.2) 0.0977

 BSE

 05/03/2003 - 0/11/2005
Trading Rule
 MA (1, 50) 9.2 * 0.1512
 MA (1, 200) (5.1) 0.0199
 MA (1, 150) (6.7) 0.1099
 Filter Rule (1%) (12.9) 0.1166
 Filter Rule (2%) (16.0) 0.1087
 Filter Rule (5%) (45.4) 0.0441
 TRB-O (50 days) 1.6 * 0.1312
 TRB-O (150 days) (20.4) 0.0770
 TRB-O (200 days) (2.4) 0.0979

 KOSPI

 20/02/2003 - 30/11/2005
Trading Rule
 MA (1, 50) (10.4) 0.0770
 MA (1, 200) (14.9) 0.0662
 MA (1, 150) (3.5) 0.0878
 Filter Rule (1%) (25.4) 0.0479
 Filter Rule (2%) (38.5) 0.0162
 Filter Rule (5%) (20.4) 0.0623
 TRB-O (50 days) (2.0) 0.0881
 TRB-O (150 days) (25.1) 0.0437
 TRB-O (200 days) (29.9) 0.0346

Chow Test (p-value) for all structural break between Sub-period
1 & 2: 0.000

Chow Test (p-value) for all structural break between Sub-period
2 & 3: 0.000

* Signifies positive returns across all three sub-periods

Table 3 - Profitability of Technical Trading Rules on Sub-Periods
(Panel B)

 Excess Sharpe Ratio
 Return

 Nikkei

 04/01/1995 - 20/8/1998
Trading Rule
 MA (1, 50) 5.8 * -0.0012
 MA (1, 200) 9.0 -0.0079
 MA (1, 150) 8.5 * 0.0052
 Filter Rule (1%) (10.0) -0.0724
 Filter Rule (2%) 0.0 -0.0377
 Filter Rule (5%) 2.1 -0.0311
 TRB-O (50 days) (3.3) -0.0347
 TRB-O (150 days) 12.4 * 0.0226
 TRB-O (200 days) 13.9 0.0146

 Straits Times

 28/12/1987 - 10/01/1994
Trading
 MA (1, 50) 6.1 0.0950
 MA (1, 200) (7.3) 0.0502
 MA (1, 150) (3.3) 0.0569
 Filter Rule (1%) 0.4 * 0.0843
 Filter Rule (2%) 0.7 * 0.0810
 Filter Rule (5%) (8.2) 0.0520
 TRB-O (50 days) 1.4 * 0.0843
 TRB-O (150 days) (6.4) 0.0423
 TRB-O (200 days) (13.6) 0.0313

 Hang Seng

 03/01/1995 - 25/9/1998
Trading Rule
 MA (1, 50) 15.9 * 0.0139
 MA (1, 200) 26.4 * 0.0198
 MA (1, 150) 22.2 0.0170
 Filter Rule (1%) 30.5 0.0446
 Filter Rule (2%) (3.5) -0.0268
 Filter Rule (5%) (2.0) -0.0234
 TRB-O (50 days) 19.8 0.0241
 TRB-O (150 days) 21.9 0.0175
 TRB-O (200 days) 25.3 0.0155

 TSEC

 02/07/1997 - 03/05/2000
Trading Rule
 MA (1, 50) 20.9 0.0277
 MA (1, 200) 10.5 0.0121
 MA (1, 150) (4.0) -0.0312
 Filter Rule (1%) 29.3 0.0428
 Filter Rule (2%) 12.5 0.0063
 Filter Rule (5%) 22.5 0.0278
 TRB-O (50 days) 10.2 0.0008
 TRB-O (150 days) 16.0 0.0146
 TRB-O (200 days) 9.4 0.0096

 Excess Sharpe Ratio
 Return

 Nikkei

 21/08/1998 - 10/04/2002
Trading Rule
 MA (1, 50) 10.0 * -0.0146
 MA (1, 200) 22.4 0.0247
 MA (1, 150) 22.5 * 0.023
 Filter Rule (1%) (8.9) -0.0722
 Filter Rule (2%) (2.5) -0.0485
 Filter Rule (5%) (3.0) -0.0539
 TRB-O (50 days) 7.9 -0.0229
 TRB-O (150 days) 16.8 * 0.0058
 TRB-O (200 days) 15.1 -0.0001

 Straits Times

 11/01/1994 - 24/12/1999
Trading
 MA (1, 50) 18.0 0.0431
 MA (1, 200) 12.4 0.0289
 MA (1, 150) 14.8 0.0347
 Filter Rule (1%) 27.0 * 0.0627
 Filter Rule (2%) 5.4 * 0.0072
 Filter Rule (5%) 15.0 0.0285
 TRB-O (50 days) 12.1 * 0.0278
 TRB-O (150 days) 14.2 0.0358
 TRB-O (200 days) 9.0 0.0218

 Hang Seng

 26/09/1998 - 05/03/2002
Trading Rule
 MA (1, 50) 3.8 * 0.0286
 MA (1, 200) 1.6 * 0.0236
 MA (1, 150) (4.6) 0.0114
 Filter Rule (1%) (6.2) 0.0082
 Filter Rule (2%) (9.7) 0.0008
 Filter Rule (5%) 22.8 0.0618
 TRB-O (50 days) (6.3) 0.0077
 TRB-O (150 days) (1.9) 0.0161
 TRB-O (200 days) (2.6) 0.0149

 TSEC

 04/05/2000 - 27/2/2003
Trading Rule
 MA (1, 50) 51.5 0.0425
 MA (1, 200) 29.6 -0.0137
 MA (1, 150) 35.0 0.0031
 Filter Rule (1%) 15.9 -0.0418
 Filter Rule (2%) 15.6 -0.0424
 Filter Rule (5%) 23.1 -0.0241
 TRB-O (50 days) 34.1 0.0001
 TRB-O (150 days) 23.0 -0.0230
 TRB-O (200 days) 13.1 -0.094

 Excess Sharpe Ratio
 Return

 Nikkei

 11/04/2002 - 20/11/2005
Trading Rule
 MA (1, 50) 8.7 * 0.0095
 MA (1, 200) (1.1) 0.0222
 MA (1, 150) 8.8 * 0.0503
 Filter Rule (1%) (8.7) 0.0000
 Filter Rule (2%) (4.1) 0.0146
 Filter Rule (5%) (10.7) -0.0065
 TRB-O (50 days) 3.3 0.0409
 TRB-O (150 days) 1.2 * 0.0247
 TRB-O (200 days) (3.0) 0.0147

 Straits Times

 25/12/1999 - 30/11/2005
Trading
 MA (1, 50) (0.4) -0.0165
 MA (1, 200) 6.7 0.0099
 MA (1, 150) 10.8 0.0267
 Filter Rule (1%) 6.1 * 0.0068
 Filter Rule (2%) 2.3 * -0.0060
 Filter Rule (5%) (3.2) -0.0231
 TRB-O (50 days) 13.4 * 0.0368
 TRB-O (150 days) 5.8 0.0059
 TRB-O (200 days) 3.2 -0.0032

 Hang Seng

 26/03/2002 - 30/11/2005
Trading Rule
 MA (1, 50) 2.1 * 0.0364
 MA (1, 200) 4.0 * 0.0428
 MA (1, 150) 2.2 0.0377
 Filter Rule (1%) (4.6) 0.0121
 Filter Rule (2%) (0.9) 0.0257
 Filter Rule (5%) (4.9) 0.0106
 TRB-O (50 days) 3.1 0.0430
 TRB-O (150 days) (3.9) 0.0148
 TRB-O (200 days) 8.1 0.0536

 TSEC

 27/02/2003 - 30/11/2005
Trading Rule
 MA (1, 50) (0.2) 0.0470
 MA (1, 200) (13.5) 0.0044
 MA (1, 150) (14.6) 0.0011
 Filter Rule (1%) (13.6) 0.0040
 Filter Rule (2%) (8.9) 0.0188
 Filter Rule (5%) (5.0) 0.0313
 TRB-O (50 days) (5.4) 0.0312
 TRB-O (150 days) (16.0) -0.0032
 TRB-O (200 days) (18.4) -0.0097

Chow Test (p-value) for all structural break between Sub-period
1 & 2: 0.000

Chow Test (p-value) for all structural break between Sub-period
2 & 3: 0.000

* Signifies positive returns across all three sub-periods.

Table 4 - Sign Prediction Ability (Panel A - one-day lag)

 MA Cross-Over Rule
 Short (days) / Long (days)

Market Index 1/50 1/200 5/150

All Ordinaries (N = 5395)
Buy Signal - Predictive Value 57.5 56.7 52.5
BPD probability of result 0.0409 * 0.1831 0.4373
Sell Signal - Predictive Value 50.7 43.3 37.5
BPD probability of result 0.4670 0.8775 0.9597

BSE (N = 2083)
Buy Signal - Predictive Value 64.4 45.8 66.7
BPD probability of result 0.0362 * 0.7294 0.1938
Sell Signal - Predictive Value 65.9 54.2 45.5
BPD probability of result 0.0244 * 0.4194 0.7256

Jakarta (N =2030)
Buy Signal - Predictive Value 69.2 62.5 42.9
BPD probability of result 0.0119 * 0.2272 0.7880
Sell Signal - Predictive Value 47.4 66.7 71.4
BPD probability of result 0.6864 0.2272 0.0898

KOSPI (N = 2067)
Buy Signal - Predictive Value 57.1 61.1 70.0
BPD probability of result 0.1958 0.2403 0.1719
Sell Signal - Predictive Value 52.1 35.3 66.7
BPD probability of result 0.4427 0.9283 0.2539

Nikkei (N = 2846)
Buy Signal - Predictive Value 43.7 33.3 28.6
BPD probability of result 0.9163 0.9739 0.9713
Sell Signal - Predictive Value 49.0 25.9 46.2
BPD probability of result 0.6167 0.997 0.7095

Hang Seng (N = 2792)
Buy Signal - Predictive Value 47.6 47.8 38.1
BPD probability of result 0.3703 0.6612 0.9054
Sell Signal - Predictive Value 47.6 34.8 33.3
BPD probability of result 0.7094 0.9534 0.9608

Straits Times (N =4479)
Buy Signal - Predictive Value 58.2 48.2 58.6
BPD probability of result 0.0346 0.6556 0.2291
Sell Signal - Predictive Value 50.0 57.1 51.7
BPD probability of result 0.5344 0.1748 0.5000

TSEC (N =2067)
Buy Signal - Predictive Value 48.8 50.0 7 57.9
BPD probability of result 0.6196 0.5700 0.3238
Sell Signal - Predictive Value 69.0 51.6 36.8
BPD probability of result 0.0098 * 0.5000 0.9165

 Filter Rule
 (in %)

Market Index 1% 2% 5%

All Ordinaries (N = 5395)
Buy Signal - Predictive Value 60.9 61.8 63.3
BPD probability of result 0.0000 * 0.0000 * 0.3506
Sell Signal - Predictive Value 51.7 47.9 55.6
BPD probability of result 0.2284 0.7544 0.1002

BSE (N = 2083)
Buy Signal - Predictive Value 59.2 57.9 59.0
BPD probability of result 0.0004 * 0.0093 * 0.1000
Sell Signal - Predictive Value 54.1 51.2 48.4
BPD probability of result 0.0804 * 0.3905 0.6482

Jakarta (N =2030)
Buy Signal - Predictive Value 56.7 61.3 58.2
BPD probability of result 0.0111 * 0.0009 * 0.0883
Sell Signal - Predictive Value 60.4 55.6 64.9
BPD probability of result 0.0003 * 0.0667 0.0070 *

KOSPI (N = 2067)
Buy Signal - Predictive Value 58.1 56.6 64.8
BPD probability of result 0.0012 * 0.0145 * 0.0013 *
Sell Signal - Predictive Value 52.5 53.3 54.8
BPD probability of result 0.1838 0.1513 0.1756

Nikkei (N = 2846)
Buy Signal - Predictive Value 45.5 44.4 43.1
BPD probability of result 0.9753 9710 0.8688
Sell Signal - Predictive Value 46.4 44.4 32.8
BPD probability of result 0.9406 0.9633 0.9973

Hang Seng (N = 2792)
Buy Signal - Predictive Value 52.5 56.3 46.8
BPD probability of result 0.1654 0.0204 * 0.7528
Sell Signal - Predictive Value 50.9 47.7 48.6
BPD probability of result 0.3825 0.7917 0.638

Straits Times (N =4479)
Buy Signal - Predictive Value 54.2 56.1 59.3
BPD probability of result 0.0262 * 0.0153 * 0.0526
Sell Signal - Predictive Value 56.9 56.4 56.5
BPD probability of result 0.0009 * 0.0144 * 0.1871

TSEC (N =2067)
Buy Signal - Predictive Value 48.2 48.7 48.4
BPD probability of result 0.0296 * 0.6770 0.6482
Sell Signal - Predictive Value 55.4 50.9 58.6
BPD probability of result 0.7719 0.422 0.0941

 Trading Range Break-Out
 (days of local max/min)

Market Index 50 150 200

All Ordinaries (N = 5395)
Buy Signal - Predictive Value 56.5 54.6 56.5
BPD probability of result 0.0104 * 0.0816 0.0323 *
Sell Signal - Predictive Value 56.1 61.1 62.8
BPD probability of result 0.1033 0.0668 0.0631

BSE (N = 2083)
Buy Signal - Predictive Value 63.1 66.7 65.6
BPD probability of result 0.0038 * 0.0026 * 0.0084 *
Sell Signal - Predictive Value 63.8 57.1 55.6
BPD probability of result 0.0240 * 0.3318 0.4073

Jakarta (N =2030)
Buy Signal - Predictive Value 65.6 69.4 67.2
BPD probability of result 0.0014 * 0.0016 * 0.0060 *
Sell Signal - Predictive Value 51.8 57.1 50.0
BPD probability of result 0.4469 0.3318 0.6047

KOSPI (N = 2067)
Buy Signal - Predictive Value 51.6 52.4 52.1
BPD probability of result 0.393 0.3703 0.4063
Sell Signal - Predictive Value 54.5 40.7 40.0
BPD probability of result 0.2693 0.8761 0.8852

Nikkei (N = 2846)
Buy Signal - Predictive Value 46.6 48.2 48.0
BPD probability of result 0.807 0.6677 0.6778
Sell Signal - Predictive Value 43.0 49.1 50.0
BPD probability of result 0.9445 0.6061 0.5551

Hang Seng (N = 2792)
Buy Signal - Predictive Value 56.0 55.2 54.4
BPD probability of result 0.0888 0.1956 0.2499
Sell Signal - Predictive Value 50.0 46.2 45.5
BPD probability of result 0.5406 0.7388 0.7566

Straits Times (N =4479)
Buy Signal - Predictive Value 61.8 57.7 58.0
BPD probability of result 0.0008 * 0.0521 0.0539
Sell Signal - Predictive Value 63.5 69.5 62.7
BPD probability of result 0.0016 * 0.0019 * 0.0460 *

TSEC (N =2067)
Buy Signal - Predictive Value 51.7 61.5 62.8
BPD probability of result 0.4152 0.0632 0.0631
Sell Signal - Predictive Value 50.0 47.4 50.0 7
BPD probability of result 5414 0.6864 0.5722

* Significant BPD probability at the 5% level

Table 4 - Sign Prediction Ability (Panel B - ten-day lag)

 MA Cross-Over Rule
 Short (days) / Long (days)

Market Index 1/50 1/200 5/150

All Ordinaries (N = 5395)
 Buy Signal - Predictive Value 60.3 61.7 47.5
 BPD probability of result 0.0080 * 0.0462 * 0.6821
 Sell Signal - Predictive Value 53.4 45.0 42.5
 BPD probability of result 0.2282 0.8169 0.8659

BSE (N = 2083)
 Buy Signal - Predictive Value 53.3 29.2 83.3
 BPD probability of result 0.383 0.9887 0.0193 *
 Sell Signal - Predictive Value 59.1 62.5 45.5
 BPD probability of result 0.1456 0.1537 0.7256

Jakarta (N =2030)
 Buy Signal - Predictive Value 67.6 46.7 57.1
 BPD probability of result 0.0235 * 0.6964 0.3953
 Sell Signal - Predictive Value 54.1 53.3 28.6
 BPD probability of result 0.3714 0.5 0.212

KOSPI (N = 2067)
 Buy Signal - Predictive Value 52.1 66.7 60.0
 BPD probability of result 0.1264 0.1189 0.377
 Sell Signal - Predictive Value 52.1 47.1 55.6
 BPD probability of result 0.4427 0.6855 0.5000

Nikkei (N = 2846)
 Buy Signal - Predictive Value 47.6 48.1 64.3
 BPD probability of result 0.5781 0.6494 0.212
 Sell Signal - Predictive Value 52.9 59.3 61.5
 BPD probability of result 0.3104 0.221 0.2905

Hang Seng (N = 2792)
 Buy Signal - Predictive Value 56.8 65.2 57.1
 BPD probability of result 0.1332 0.105 0.3318
 Sell Signal - Predictive Value 47.6 56.5 47.6
 BPD probability of result 0.7094 0.3388 0.6682

Straits Times (N =4479)
 Buy Signal - Predictive Value 47.7 42.9 51.7
 BPD probability of result 0.7287 0.8856 0.5
 Sell Signal - Predictive Value 51.1 51.8 62.1
 BPD probability of result 0.4312 0.4469 0.1325

TSEC (N =2067)
 Buy Signal - Predictive Value 65.1 56.7 38.9
 BPD probability of result 0.0330 * 0.2923 0.8811
 Sell Signal - Predictive Value 57.1 43.3 36.8
 BPD probability of result 0.2204 0.4278 0.9165

 Filter Rule
 (in %)

Market Index 1% 2% 5%

All Ordinaries (N = 5395)
 Buy Signal - Predictive Value 62.2 59.7 73.3
 BPD probability of result 0.0000 * 0.0006 * 0.9739
 Sell Signal - Predictive Value 48.8 45.0 33.3
 BPD probability of result 0.7302 0.9352 0.0081 *

BSE (N = 2083)
 Buy Signal - Predictive Value 55 55.6 63.3
 BPD probability of result 0.0367 * 0.0051 * 0.0259 *
 Sell Signal - Predictive Value 47.8 50.5 50.0
 BPD probability of result 0.8006 0.4722 0.5505

Jakarta (N =2030)
 Buy Signal - Predictive Value 56.6 58.4 59.5
 BPD probability of result 0.0126 * 0.0112 * 0.0573
 Sell Signal - Predictive Value 52.9 50.5 45.9
 BPD probability of result 0.1842 0.4715 0.792

KOSPI (N = 2067)
 Buy Signal - Predictive Value 55.2 55.7 58.3
 BPD probability of result 0.0262 * 0.0294 * 0.0507 *
 Sell Signal - Predictive Value 47.2 48.0 50.4
 BPD probability of result 0.8678 07670 0.5000

Nikkei (N = 2846)
 Buy Signal - Predictive Value 49.6 50.6 60.0
 BPD probability of result 0.5925 0.4513 0.1013
 Sell Signal - Predictive Value 50.7 48.4 34.5
 BPD probability of result 0.4077 0.7246 0.994

Hang Seng (N = 2792)
 Buy Signal - Predictive Value 53.1 55.4 55.8
 BPD probability of result 0.1118 0.0403 * 0.181
 Sell Signal - Predictive Value 47.8 46.1 51.4
 BPD probability of result 0.8283 0.9054 0.4531

Straits Times (N =4479)
 Buy Signal - Predictive Value 54.7 53.5 67.4
 BPD probability of result 0.0140 * 0.1111 0.0008 *
 Sell Signal - Predictive Value 49.4 46.9 40.3
 BPD probability of result 0.6208 0.8747 0.951

TSEC (N =2067)
 Buy Signal - Predictive Value 49.9 51.3 61.3
 BPD probability of result 0.2706 0.3714 0.0490 *
 Sell Signal - Predictive Value 51.9 53.9 54.3
 BPD probability of result 0.5426 0.1322 0.2752

 Trading Range Break-Out
 (days of local max/min)

Market Index 50 150 200

All Ordinaries (N = 5395)
 Buy Signal - Predictive Value 63.1 63.5 64.0
 BPD probability of result 0.0000 * 0.0000 * 0.0000 *
 Sell Signal - Predictive Value 48.8 55.6 53.5
 BPD probability of result 0.6407 0.2383 0.3804

BSE (N = 2083)
 Buy Signal - Predictive Value 67.3 64.9 65.1
 BPD probability of result 0.0003 * 0.0070 * 0.0113 *
 Sell Signal - Predictive Value 46.6 38.1 38.9
 BPD probability of result 0.7441 0.9054 0.8811

Jakarta (N =2030)
 Buy Signal - Predictive Value 60.4 56.5 53.4
 BPD probability of result 0.0260 * 0.1871 0.347
 Sell Signal - Predictive Value 51.8 76.2 64.3
 BPD probability of result 0.4469 0.0133 * 0.212

KOSPI (N = 2067)
 Buy Signal - Predictive Value 62.0 61.7 58.6
 BPD probability of result 0.0053 * 0.0224 * 0.0941
 Sell Signal - Predictive Value 51.5 29.6 24.0
 BPD probability of result 0.4511 0.9904 0.998

Nikkei (N = 2846)
 Buy Signal - Predictive Value 58.3 58.3 58.1
 BPD probability of result 0.0413 * 0.0778 0.1003
 Sell Signal - Predictive Value 44.7 52.7 50.0
 BPD probability of result 0.8884 0.3939 0.5551

Hang Seng (N = 2792)
 Buy Signal - Predictive Value 56.0 59.8 59.5
 BPD probability of result 0.0888 0.0428 * 0.0573
 Sell Signal - Predictive Value 52.1 48.7 48.5
 BPD probability of result 0.3798 0.6254 0.6358

Straits Times (N =4479)
 Buy Signal - Predictive Value 54.8 52.8 54.5
 BPD probability of result 0.1062 0.2943 0.1976
 Sell Signal - Predictive Value 50.8 49.2 47.1
 BPD probability of result 0.4645 0.6026 0.7121

TSEC (N =2067)
 Buy Signal - Predictive Value 55.8 50.0 55.8
 BPD probability of result 0.1659 0.5551 0.2712
 Sell Signal - Predictive Value 46.7 50.0 46.7
 BPD probability of result 0.986 0.5643 0.7077

* Significant BPD probability at the 5% level.

Table 5: Sign Prediction Ability (Panel A - one-day lag)

 MA Cross-Over Rule
 Short (days) / Long (days)

Market Index 1/50 1/200 5/150

All Ordinaries (N = 5395)

Daily Ave. % Return after Signal (Lag 1)
 Buy Signal 0.0006 0.0020 0.0005
 Sell Signal 0.0000 0.0008 0.0006
 Buy-Sell t-stat 0.4888 0.7572 -0.0552

BSE (N = 2083)
 Buy Signal 0.0039 -0.0019 0.0003
 Sell Signal -0.0071 -0.0060 -0.0019
 Buy-Sell t-stat 3.2468 * 0.5971 0.2729

Jakarta (N =2030)
 Buy Signal 0.0072 0.0081 -0.0022
 Sell Signal 0.0000 0.0023 -0.0130
 Buy-Sell t-stat 1.9924 * 0.8866 0.9871

KOSPI (N = 2067)
 Buy Signal 0.0064 -0.0036 0,0130
 Sell Signal -0.0027 0.0015 -0.0082
 Buy-Sell t-stat 2.2761 * -0.8318 1.7506

Nikkei (N = 2846)
 Buy Signal -0.0003 -0.0049 0.0006
 Sell Signal 0.0000 0.0065 -0.0009
 Buy-Sell t-stat -0.1664 -4.1577 0.2398

Hang Seng (N = 2792)
 Buy Signal 0.0009 -0.0023 -0.0029
 Sell Signal -0.0011 -0.0001 0.0028
 Buy-Sell t-stat 0.9181 -0.4610 -1.6292

Straits Times (N =4479)
 Buy Signal 0.0023 -0.0029 0.0018
 Sell Signal -0.0005 -0.0026 -0.0047
 Buy-Sell t-stat 1.9645 * -0.1287 1.9659 *

TSEC (N =2067)
 Buy Signal 0.0019 0.0025 0.0035
 Sell Signal -0.0066 -0.0008 0.0014
 Buy-Sell t-stat 3.3046 * 0.7497 0.4027

 Filter Rule
 (in %)

Market Index 1% 2% 5%

All Ordinaries (N = 5395)

Daily Ave. % Return after Signal (Lag 1)
 Buy Signal 0.0022 0.0031 -0.0014
 Sell Signal -0.0007 -0.0017 -0.0125
 Buy-Sell t-stat 5.4119 * 1.9464 0.9240

BSE (N = 2083)
 Buy Signal 0.0013 0.0022 0.0033
 Sell Signal -0.0022 -0.0022 -0.0029
 Buy-Sell t-stat 2.6735 * 2.6783 * 1.4259

Jakarta (N =2030)
 Buy Signal 0.0005 0.0059 0.0079
 Sell Signal -0.0059 -0.0044 -0.0084
 Buy-Sell t-stat 6.8629 * 4.9677 * 3.9906 *

KOSPI (N = 2067)
 Buy Signal 0.0029 0.0020 0.0080
 Sell Signal -0.0029 -0.0033 -0.0051
 Buy-Sell t-stat 3.1819 * 2.5313 * 3.4929 *

Nikkei (N = 2846)
 Buy Signal -0.0008 -0.0008 -0.0007
 Sell Signal -0.0001 0.0010 0.0063
 Buy-Sell t-stat -0.6461 -1.4785 -1.8289

Hang Seng (N = 2792)
 Buy Signal 0.0025 0.0041 0.0002
 Sell Signal -0.0018 -0.0020 -0.0035
 Buy-Sell t-stat 3.4883 * 3.4178 * 0.9075

Straits Times (N =4479)
 Buy Signal 0.0017 0.0024 0.0055
 Sell Signal -0.0027 -0.0025 -0.0038
 Buy-Sell t-stat 5.5074 * 3.8112 * 2.6567 *

TSEC (N =2067)
 Buy Signal 0.0003 0.0005 0.0026
 Sell Signal -0.0015 -0.0017 -0.0045
 Buy-Sell t-stat 1.3831 1.3933 2.1690 *

 Trading Range Break-Out
 (days of local max/min)

Market Index 50 150 200

All Ordinaries (N = 5395)

Daily Ave. % Return after Signal (Lag 1)
 Buy Signal 0.0014 0.0012 0.0013
 Sell Signal -0.0053 -0.0040 -0.0044
 Buy-Sell t-stat 2.5672 * 2.1105 * 1.9397

BSE (N = 2083)
 Buy Signal 0.0035 0.0044 0.0037
 Sell Signal -0.0080 0.0001 -0.0040
 Buy-Sell t-stat 3.4970 * 0.6865 1.3630

Jakarta (N =2030)
 Buy Signal 0.0050 0.0066 0.0049
 Sell Signal -0.0005 -0.0048 0.0002
 Buy-Sell t-stat 1.5182 2.2995 * 2.2112 *

KOSPI (N = 2067)
 Buy Signal 0.0009 0.0022 -0.0000
 Sell Signal -0.0050 0.0074 0.0085
 Buy-Sell t-stat 1.3296 -0.7881 -1.2650

Nikkei (N = 2846)
 Buy Signal 0.0003 0.0003 0.0003
 Sell Signal 0.0008 0.0004 0.0008
 Buy-Sell t-stat -0.2414 -0.0373 -0.1796

Hang Seng (N = 2792)
 Buy Signal 0.0031 0.0031 0.0029
 Sell Signal 0.0003 0.0025 0.0035
 Buy-Sell t-stat 0.9652 0.08636 -0.0779

Straits Times (N =4479)
 Buy Signal 0.0035 0.0033 0.0033
 Sell Signal -0.0044 -0.0067 -0.0053
 Buy-Sell t-stat 4.7639 * 3.9465 * 3.1754 *

TSEC (N =2067)
 Buy Signal -0.0003 0.0006 0.0008
 Sell Signal -0.0003 -0.0003 -0.0031
 Buy-Sell t-stat -0.0090 0.2170 0.8538

The t-stat critical values are as follows: 1.645 at 0.10 [alpha],
1.96 at the 0.05 [alpha], and 2.576 at the [alpha]. * Significant
p-values at the 5% level.

Table 5 - Sign Prediction Ability (Panel B - ten-day lag)

 MA Cross-Over Rule
 Short (days) / Long (days)

Market Index 1/50 1/200 5/150

All Ordinaries (N = 5395)

Daily Ave. % Return after Signal (Lag 10)
 Buy Signal 0.0048 0.0044 0.0010
 Sell Signal -0.0028 0.0016 -0.0051
 Buy-Sell t-stat 1.5349 0.6576 0.7946

BSE (N = 2083)
 Buy Signal -0.0039 -0.0300 0.0079
 Sell Signal -0.0218 -0.0298 -0.0154
 Buy-Sell t-stat 1.3394 -0.0112 0.9041

Jakarta (N =2030)
 Buy Signal 0.0287 -0.0046 0.0016
 Sell Signal -0.0062 -0.0141 -0.0071
 Buy-Sell t-stat 2.2836 * 0.4720 1.0526

KOSPI (N = 2067)
 Buy Signal 0.0063 0.0116 0.0156
 Sell Signal -0.0121 -0.0100 -0.0009
 Buy-Sell t-stat 1.2038 0.9851 0.5593

Nikkei (N = 2846)
 Buy Signal -0.0010 -0.0064 0.0163
 Sell Signal -0.0021 -0.0071 -0.0096
 Buy-Sell t-stat 0.1995 0.0829 1.9132

Hang Seng (N = 2792)
 Buy Signal 0.0013 0.0033 0.0044
 Sell Signal -0.0019 -0.0190 -0.0057
 Buy-Sell t-stat 0.3594 0.9797 0.6309

Straits Times (N =4479)
 Buy Signal -0.0038 0.0050 0.0063
 Sell Signal -0.0045 0.0043 -0.0100
 Buy-Sell t-stat 0.1585 0.1056 1.7652

TSEC (N =2067)
 Buy Signal 0.0132 -0.0035 0.0046
 Sell Signal -0.0048 -0.0012 0.0154
 Buy-Sell t-stat 1.8162 -0.1616 -0.5341

 Filter Rule
 (in %)

Market Index 1% 2% 5%

All Ordinaries (N = 5395)

Daily Ave. % Return after Signal (Lag 10)
 Buy Signal 0.0065 0.0091 0.0108
 Sell Signal -0.0001 -0.0005 -0.0063
 Buy-Sell t-stat 2.9349 * 1.2567 0.7188

BSE (N = 2083)
 Buy Signal 0.0039 0.0029 0.0058
 Sell Signal -0.0016 -0.0052 -0.0045
 Buy-Sell t-stat 1.2741 1.9049 0.9388

Jakarta (N =2030)
 Buy Signal 0.0073 0.0088 0.0162
 Sell Signal -0.0075 -0.0050 0.0038
 Buy-Sell t-stat 2.5903 * 1.8548 0.9788

KOSPI (N = 2067)
 Buy Signal 0.0005 0.0018 0.0117
 Sell Signal -0.0024 0.0013 -0.0003
 Buy-Sell t-stat 0.5175 0.0830 1.1708

Nikkei (N = 2846)
 Buy Signal -0.0030 -0.0016 0.0070
 Sell Signal -0.0010 0.0027 0.0200
 Buy-Sell t-stat -0.7013 -1.1342 -1.5784

Hang Seng (N = 2792)
 Buy Signal 0.0040 0.0058 0.0046
 Sell Signal 0.0030 0.0037 0.0008
 Buy-Sell t-stat 0.2415 0.3407 0.3327

Straits Times (N =4479)
 Buy Signal 0.0059 0.0066 0.0168
 Sell Signal -0.0013 0.0000 0.0079
 Buy-Sell t-stat 2.4111 * 1.5404 0.8592

TSEC (N =2067)
 Buy Signal -0.0009 -0.0027 0.0117
 Sell Signal -0.0050 -0.0052 -0.0002
 Buy-Sell t-stat 0.8791 0.4409 1.0968

 Trading Range Break-Out
 (days of local max/min)

Market Index 50 150 200

All Ordinaries (N = 5395)

Daily Ave. % Return after Signal (Lag 10)
 Buy Signal 0.0059 0.0059 0.0064
 Sell Signal -0.0046 -0.0049 -0.0003
 Buy-Sell t-stat 2.2001 * 7.5027 * 1.0402

BSE (N = 2083)
 Buy Signal 0.0126 0.0127 0.0128
 Sell Signal -0.0035 0.0103 0.0099
 Buy-Sell t-stat 2.0400 * 0.1630 0.1741

Jakarta (N =2030)
 Buy Signal 0.0132 0.0077 0.0061
 Sell Signal -0.0177 -0.0480 -0.0399
 Buy-Sell t-stat 2.7074 * 3.1290 * 1.9925 *

KOSPI (N = 2067)
 Buy Signal 0.0056 0.0057 0.0013
 Sell Signal 0.0058 0.0454 0.053
 Buy-Sell t-stat -0.0236 -2.5633 -3.3048

Nikkei (N = 2846)
 Buy Signal -0.0007 -0.0039 -0.0033
 Sell Signal 0.0070 -0.0007 0.0015
 Buy-Sell t-stat -1.4223 -0.3925 -0.5577

Hang Seng (N = 2792)
 Buy Signal 0.0042 0.0056 0.0062
 Sell Signal 0.0033 0.0116 0.0133
 Buy-Sell t-stat 0.1142 -0.5158 -0.4985

Straits Times (N =4479)
 Buy Signal 0.0091 0.0065 0.0082
 Sell Signal -0.0072 -0.0056 -0.0029
 Buy-Sell t-stat 2.7428 * 1.3438 1.1579

TSEC (N =2067)
 Buy Signal 0.0060 0.0005 0.0048
 Sell Signal -0.0008 0.0028 -0.0034
 Buy-Sell t-stat 0.8330 0.2449 0.5145

The t-stat critical values are: 1.645 at 0.10 [alpha], 1.96 at the
0.05 [alpha], and 2.576 at the [alpha]. * Significant p-values at
the 5% level.