Skip to main Content
Site Search

Advanced Search

  • Mondo Visione
  • Mondo Visione - Worldwide Exchange Intelligence
Member Login

Member Login

Forgotten your password?

Using technical analysis to test the efficiency of the Shanghai Stock Market

Date 29/10/2008

Andrea Coppola and Juan Zhang
University of Rome

Can technical analysis beat the market? Numerous financial economists and practitioners have analysed this issue, but a shared conclusion has not yet been reached.

The great majority of academics support the random walk hypothesis and the well-known efficient market hypothesis (EMH) developed by Fama (1970). Under the assumption of weak form efficiency, security prices accurately reflect all security market information including past price changes and trading activities. Consequently, any statistical analysis based on past price and volume information should not have any predictive power on the future performance of the stocks. On the other hand, significant publications recognise that ‘technical analysis may well be an effective means for extracting useful information from market prices’ (Lo et al., 2000). In contrast to the efficient market hypothesis, many recent studies showed that the application of technical analysis can in fact generate excess returns.

Among these studies, the ‘CRISMA’ trading system (Pruitt and White, 1988) has drawn particular attention. CRISMA is a momentum-based system which seeks to identify relatively good performance by a security in the expectation that such performance will continue. CRISMA combines three of the most common indicators used by technical analysts: moving averages, trading volumes and relative strengths. The work of Pruitt and White showed that the strategy suggested by this multiple filter system could produce significant profits in US stock market.

This article sets out to evaluate the profitability of the CRISMA system in the context of the Shanghai Stock Exchange (SSE). We chose the largest Chinese stock market – and the fifth largest in the world – for a number of reasons. China's rapid economic growth and the development of its financial markets mutually reinforce each other. However, the peculiarity of the Chinese political end economic system could raise doubts about the efficiency of its stock markets (Chan et al., 2007). Moreover, despite the increasing economic and financial importance of this huge country, Chinese stock markets have received little research attention. Indeed, there have been very few studies concerning the profitability of technical trading rules based on the data of any Asian stock markets. We used the CRISMA trading system to test the efficiency of the Shanghai stock market over the period from June 1996 to July 2005. Consistently with existing literature, our results confirm the efficiency of the A-market (where shares are denominated in Renminbi) but cast some doubts on the efficiency of the B-market (where shares are denominated in US dollars).

The value of technical analysis

Technical analysis is probably the main source of discord between finance academics and practitioners. The usefulness of technical analysis has been extensively debated in the last decades and the literature is rich with studies which investigate its value.

The most interesting outcome from these debates concerns the contrast between the use of technical analysis and the market efficiency hypothesis. If markets are efficient, technical analysis should have no contribution to make. If technical analysis works, markets cannot be efficient. In fact, many studies have confirmed the predictability of technical trading rules, both when focusing on developed markets (Pruitt and White, 1988; Sweeney, 1988; Brock et al., 1992; Corrado and Lee, 1992; Pruitt et al., 1992; Hinkel et al., 1996) and when considering developing markets (Huang, 1995; Wong, 1995; Raj and Thurston, 1996; Wong et al., 2003).

Most of these studies relied on simple trading strategies founded on the main technical analysis evaluation tools: moving averages, trading volumes and relative strengths. Moving average is the indicator most commonly used by technical analysts. It focuses on stock performance trends and in particular helps to identify two different kinds of trend: long-term trends and short-term ones. The identification of buying and selling signals is based on the joint evaluation of these trends. Stocks are purchased (sold) if the short-term moving average rises above (below) the long-term average, since the crossing of these two moving averages is considered to be the beginning of a trend (positive or negative, respectively) for the stock under consideration. In addition to simplicity, another reason for the popularity of moving averages is that they smooth out market fluctuations and short-term volatility. With regard to the profitability of this approach, earlier studies provide mixed evidence. The results of Cootner (1962), based on NYSE stocks between 1956 and 1960, were in favour of moving average rules, whereas Van Horne and Parker (1967, 1968), where NYSE is considered during the period 1960-1966, concluded that a simple buy-and-hold strategy performs better than moving averages. Among more recent studies, Brock et al. (1992) tested the profitability of different kinds of moving average by using 90 years of data from the Dow Jones Industrial Average (from 1897 to 1985). Their results confirmed the profitability of moving averages and the authors stated that ‘the conclusion reached by many earlier studies that found technical analysis to be useless might have been immature’. A similar analysis by Hinkel et al. (1996) which focused on the UK market from 1935 to 1994 confirmed the predictive power of technical trading rules. However, they also argued that these rules are not profitable when taking into account transaction costs. Therefore, they inferred that weak form market efficiency is not violated.

Trading volume is another variable widely considered in technical analysis. Rogalski (1978) suggested that knowledge of the behaviour of volume may marginally improve the conditional price forecasts compared to price forecasts based on past prices alone. Karpoff’s (1987) survey of the relationship between price changes and trading volume found that 12 out of 16 studies considered support a positive correlation. More recently, Hiemstra and Jones (1994) found evidence of bi-directional causality between returns and volume in the US market, as did Silvapulle and Choi (1997) in the Korean market. Saatcioglu and Starks (1998) investigated the relationship between stock prices and traded volumes in Latin America emerging stock markets and found that traded volumes led stock price changes in four of the six markets considered.

Technical analysts also take into account the relative strength of stocks which measures the price performance of a stock versus a market average or universe of stocks. A stock's relative strength improves if the stock rises more than the market in an uptrend, or goes down less than the market in a downtrend. Given a certain market environment, securities with higher relative strength perform better than others and this behaviour is presumed to continue over time. Levy (1967) confirmed this hypothesis by considering NYSE data between 1960 and 1965. More recently O'Shaughnessy (2005), who tested a large set of trading strategies in the US market over a period from 1951 to 1996, showed that relative strength is one of the criteria in all ten of the top-performing strategies.

In recent years, trading strategies have evolved to consider the joint effect of these technical indicators. Wong et al. (2003) showed that moving average and relative strength indicators could be used to generate significantly positive returns on the Singapore Stock Exchange between 1974 and 1994. Pruitt and White (1988) proposed the CRISMA system, a three-component trading system based on cumulative volumes, relative strengths and moving averages, which showed encouraging results for the US stock market (sample period: 1976-1985) even after taking into account transaction costs. Some years later, Pruitt et al. (1992) considered a different sample period (1986-1990) to confirm the profitability of the CRISMA system in the US market. However, Goodacre et al. (1999) and Goodacre and Kohn-Speyer (2001) re-examined the CRISMA system in the UK and US stock markets (sample period: 1988-1996) and found no evidence of profitability when returns were adjusted for market movements. Cheng et al. (2003) tested CRISMA in the Hong Kong stock market and found some evidence of profitability between 1990 and 2001. They stated that Asian stock markets are not as efficient as other well-developed stock markets and hence could be exploited by technical analysis.

In our study we performed the CRISMA analysis at the Shanghai Stock Exchange, the first attempt to use technical analysis to test for efficiency in the Chinese stock markets. Previously, Chinese market efficiency has been tested using different approaches. Cai et al. (1997) and Groenewold et al. (2003) (sample period: 1993-1996 and 1992-2001, respectively) used serial correlation tests. Both found that the domestic Chinese A-share market is weak-form efficient, while the B-share market is not. Barnes and Ma (2001) and Lock (2007) computed variance ratio test statistics (sample period: 1990-1998 and 1992-2007, respectively). In line with previous research, they found that the B-share market is more predictable than the A-share market.

Data and methodology

Data sample

Due to impressive growth in recent years, the Shanghai Stock Exchange (SSE) has overtaken the Hong Kong Stock Exchange (HKSE) and, with a market capitalisation of USD2.4tr, is far larger than the other mainland Chinese stock exchange, located in Shenzhen (market capitalisation USD0.7tr). Two types of stock are traded in the SSE: A-shares and B-shares. A-shares are priced in Renminbi while B-shares are quoted in US dollars. Initially, trading in A-shares was restricted to domestic investors while B-shares were available to foreign investors only but currently the B-market is also open to domestic investors while foreign investors are allowed (with limitations) to trade in A-shares under the Qualified Foreign Institutional Investor (QFII) system. Stocks on SSE are traded from Monday to Friday. The morning session from 09:15 to 11:30 is followed by the afternoon consecutive bidding session, from 13:00 to 15:00.

Our analysis separately evaluated the A-shares market (824 stocks) and B-shares market (54 stocks) to test whether there was any difference in terms of market efficiency. The data, from Datastream, covered the period from June 1996 to July 2005 and comprised daily closing prices (P) and turnover by volume (VO), which measures the number of shares traded for a particular stock on a particular day. Share prices and trading volumes were adjusted for capital actions (such as stock splits and stock dividends). Shanghai A-share and B-share price indices (PI) were used as a benchmark to evaluate shares listed in the A-market and B-market, respectively.

The CRISMA trading system

Shanghai stock market efficiency was tested by evaluating the profitability of the CRISMA trading rule. CRISMA analysis is based on past volume and price information. If the market is efficient, CRISMA analysis should not be able to produce excess returns.

Equity targets are selected by CRISMA through a filter system which picks out the upward momenta of the stocks considered. In particular, three filters are used to evaluate stock performance over time. The first filter is based on stock price moving averages (200 days and 50 days moving averages) and compares the long-term trend with the short-term one. The filter is satisfied when the 200 days moving average is not decreasing and, at the same time, the 50 days moving average crosses the 200 days moving average from below. In technical analysis jargon this is called a ‘golden cross’ and it indicates when the short-term performance of a stock outstrips its long-term performance. The second filter is based on volume information. The cumulative volume for each stock is computed by adding traded volumes when the price has increased with respect to the previous day and subtracting them when the price has decreased with respect to the previous day. In other words, cumulative volume provides information about the demand for the stock. The filter is satisfied when the cumulative volume has grown over the last four weeks considered, i.e. when the demand for the stock has increased. The third filter concerns the relative strength, which is defined as the ratio between the stock price and the market index. This indicator focuses on the performance of the stock considered compared to that of the market as a whole. The filter is satisfied when, over the last four weeks, the relative strength of the stock has increased.

When the upward momentum of a stock is confirmed on the same day by all three filters, the CRISMA trading rule applies a final confirmation filter to reduce the likelihood of false signals. With this aim, the stock is purchased when its price reaches 110% of the intersection point between the 50 and 200 day price moving averages, while the purchased stock is sold when either its price rises above 120% of the intersection point or declines below the 200 day moving average value. According to the most recent literature on CRISMA (Cheng et al., 2003; Goodacre and Kohn-Speyer, 2001), these general rules are completed by the following exceptions: (i) if the price of the stock does not reach 110% of the intersection point between the moving averages within five weeks of the first three filters being satisfied, it is not purchased; (ii) if the stock price is already above the 110% level when the first three filters are satisfied, it is not purchased until it falls below this level, and (iii) the stock is not purchased if its price is already above the 110% level when the first three filters are satisfied and reaches 120% of the intersection point of the moving averages without falling below the 110% level.

In order to illustrate the procedure described, we focus on the first stock considered in our data set, namely the shares issued by the Shanghai Pudong Development Bank (SPDB). SPDB is a commercial bank which provides financial services for the development of the region of Pudong and the city of Shanghai. It was listed on the Shanghai Stock Exchange on 10 November 1999. By performing the CRISMA analysis on SPDB shares, we identified clear trading signals. In particular, the first three filters were satisfied on 3 July 2002.

Figure 1: Moving averages of SPDB shares, 2002

Figure 1 shows that between the 2nd and the 3rd of July, the 50 day moving average computed from SPDB share prices crossed the 200 day moving average from below. At the same time, the 200 day moving average did not decrease. Given these conditions, the intersection point can be defined as a ‘golden cross’ and the first filter is satisfied.

Figure 2: Cumulative volume of SPDB shares, 2002

In Figure 2,. the dotted line, which connects the cumulative volume value computed for the ‘golden cross’ day and its value four weeks before, is positively sloped. This shows that the demand for the stock increased, which satisfies the second filter.

Figure 3: Relative strength of SPDB shares, 2002

The dotted line in Figure 3 shows that the relative strength of the stock increased over the four weeks preceding the ‘golden cross’; hence the third filter is also satisfied.

Figure 4: The chart analysis allows to identify trading signals

Since the first three filters are satisfied, we compute the values at 110% (CNY11.40) and 120% (CNY12.44) of the intersection point between the moving averages and focus on share price dynamic (see Figure 4). Given that the stock price on 3 July 2002 (CNY11.79) was already above the 110% level, the exceptions to the general CRISMA rules have to be considered. In particular, if the price of the stock is already above the 110% confirmation filter when the first three filters are satisfied, the stock is not purchased until it falls below this 110% level. The price went below the 110% level on 10 July 2002 (CNY11.21). That is the buying signal, triggering purchase of the stock on the following day. The length of the holding period also depends on price dynamics. The stock is not sold until either the price overtakes the 120% level or goes below the 200 day moving average. Figure 4 shows that the 120% price level was reached on 23 August 2002. That is the selling signal, triggering sale of the stock on the following working day. Overall, the holding period for the stock was 32 working days. Given the buying price (11.21) and the selling price (12.77), the gross return from this trade was 13.9%. Unfortunately, the CRISMA trading rule does not always provide this level of profitability.

The CRISMA trading system in the Shanghai Stock Market

Empirical results

Following the CRISMA trading rule, a total number of 594 trades in A-shares market and 32 trades in B-shares market were identified. These trades concerned 397 companies in the A-share market and 24 companies in the B-share market. The total holding period for trades identified in the A-market was 14,384 security-days while for the B-market it was 502 security-days.


Table 1: Gross returns from CRISMA trading system, June 1996 to July 2005

Transaction costs (%) A-share market - Mean returns per trade (%) Annualised returns (%)
0.0 0.532 5.49
0.5 0.32 0.33
1.0 -0.468 -4.83
1.5 -0.968** -9.99
2.0 -1.47*** -15.16
Transaction costs (%) B-share market - Mean returns per trade (%) Annualised returns (%)
0.0 3.84* 61.24
0.5 3.34* 53.28
1.0 2.84 45.31
1.5 2.34 37.34
2.0 1.84 29.37
Note: Results are tested to evaluate whether mean returns per trade are significantly positive or negative. One, two and three asterisks stand for the rejection of the null hypothesis at the 10% level, at the 5% level and at the 1% level, respectively.

Table 1 presents statistics about the gross profits generated by the CRISMA system, assuming different levels of transaction cost. When considering the A-share market, the mean gross return per trade is positive but very low. Even if transaction costs are not considered, the figure is not statistically significant. After allowing for round-trip transaction costs of up to 2%, losses become statistically significant. According to the level of transaction costs assumed, arithmetically annualised returns (based on a 250 business day year) ranged from 5.5% to -15.2%.

The outcome of the analysis is somewhat different when we focus on the B-share market, where the CRISMA system seems to show some profitability: a 3.8% mean gross profit per trade, and returns which are still positive and statistically significant even allowing for (small) transaction costs. Annualised profits range from 61.2% to 29.4% depending on the level of transaction costs assumed.

Overall, the CRISMA trading system does not show any profitability when applied to the A-share market. On the other hand, the CRISMA system casts doubts on the efficiency of Shanghai Stock Exchange when considering the B-share market.

So far, the figures presented just refer to gross return results. In order to fully assess the profitability of the trading system, the returns were adjusted taking into account market movements. In particular, excess returns were computed by using two different return generating models: the mean adjusted model and the market adjusted model.

Expectations in the mean adjusted model are measured by computing, for each security considered, the simple average of daily returns over a control period of 200 days prior to the security being purchased. Excess return equals the difference between the return produced by the CRISMA trading system and the performance of the security in the period preceding the purchase.

This approach provides further evidence of efficiency for the SSE A-share market. Even if transaction costs were not taken into account, the mean daily excess returns produced by CRISMA were not statistically significant. Similar results were obtained when considering the B-share market: mean daily excess returns were generally positive but not statistically significant at conventional levels.

However, some literature (Goodacre et al., 1999) points out that mean adjusted excess returns could underestimate the performance of the CRISMA system in the sense that the return generating model considered could have some good performance built into the expectations. To control for this potential problem, a market adjusted excess return model was considered which compares trade returns with the return of the whole market.

But despite the different approach, the conclusion did not change. Excess returns for both the A-share market and B-share market were not statistically different from zero.

The remarkable difference between the gross returns provided by the CRISMA trading system and the returns adjusted for market movements seems to confirm the importance of focusing on excess returns when evaluating the profitability of the CRISMA system and, consequently, the efficiency of Shanghai Stock Market. Gross returns could be biased by the general performance of the financial markets: if the whole market performs well in a certain period. Nevertheless, if CRISMA, given past information about prices and volumes, is able to identify the right moment to purchase a stock, the trading system should be considered profitable, no matter what the relative performance of the market in the same period.

In order to further control the robustness of the results obtained, a binomial proportionality test was performed to test statistically whether the CRISMA system predict more profitable than unprofitable trades.

Statistics were computed for both gross and adjusted returns. Overall, the CRISMA system was not able to predict more profitable than unprofitable trades in the A-share market and taking into account transaction costs, the performance of the trading rule was significantly negative. From these outcomes, the efficiency of the A-share market is again confirmed. Results for the B-share market provide some weak evidence about the profitability of the trading system but overall, results from the binomial proportionality test are consistent with what was found while focusing just on returns.

The 2001-2002 CSRC reform

In order to enrich the empirical analysis and assess the market efficiency of the Shanghai Stock Exchange, the reform which changed the structure of Chinese stock markets was considered. Until this reform, trading in A-shares was restricted to domestic investors while B-shares were only available to foreign investors. The aim of the reform was to reduce the market fragmentation of Chinese stock markets which was achieved by opening the B-market to domestic investors (19th February 2001) and allowing some qualified foreign investors to trade in the A-market (1 December 2002).

To take into account the effects of this reform, the sample period was split into two sub-samples: for the A-market, 1 June 1996 to 30 November 2002 and 1 December 2002 to 31 July 2005; for the B-market, 1 June 1996 to 18 February 2001, and 19 February 2001 to 31 July 2005.

Results of the analysis confirm the findings obtained for the A-market and shed light on the issue of B-market efficiency. When considering the A-market, results obtained from the two different sub-samples were homogeneous and consistent with the outcomes of the exercise performed on the whole sample. The evidence supports the efficiency of the A-market since the CRISMA system was not able to produce excess returns and did not have any market timing ability. When considering the B-market, the divergence between the results from the two sub-samples highlighted the effect of reform on the B-market and clarified that B-market inefficiency was driven by the restrictions on trading which characterised the pre-2001 period. When focusing on the first sub-sample, gross returns were positive and statistically significant at the 5% level, even after allowing for transactions costs. The evidence is even more striking when considering the proportion of profitable trades, which is statistically significant at the 1% level for both gross returns and mean adjusted excess returns. On the other hand, results obtained for the second sub-sample show that the CRISMA trading system was not profitable in the post-2001 period.

Conclusion

‘Who says technical analysis can't beat the market?’ This provoking statement concludes Pruitt and White's article (1988), in which the CRISMA trading system is shown to be able to exploit information about past prices and trading volumes to predict US stock market movements.

The CRISMA trading system did not show any profitability when applied to the Shanghai A-share market. This evidence supports Shanghai A-market efficiency. On the other hand, small but significant achievements of the CRISMA system cast doubt on the efficiency of Shanghai B-market (in line with the existing literature). These results suggested that the size of the market could be important in determining market efficiency as a smaller size (the B-market is much smaller than the A-market) might imply lower efficiency. However, by splitting the sample considered, this study suggests that B-market inefficiency was greatest in the pre-2001 reform period.

This underlines the fundamental role played by regulators. In spite of the rapid growth which characterised Chinese financial markets during the 1990s, the B-market became efficient only after the 2001 CSRC reform. The indirect but interesting link between the outcome of technical analysis and the effectiveness of policy interventions could be an object of future research. While focusing on developing markets, the assessment of technical analysis profitability could be a useful indicator to evaluate the impact of regulators’intervention.

References

Brock, W, Lakonishok, J, & LeBaron, B (1992). Simple technical trading rules and the stochastic properties of stock returns. Journal of Finance, 47, 1731-64.

Brown, S J, & Warner, J B (1985). Using daily stock returns: The case of event studies. Journal of Financial Economics, 14, 3-31.

Cai, F, Laurence, M, & Qian, S (1997). Weak-form efficiency and causality tests in Chinese Stock markets. Multinational Finance Journal, 1, 291-307.

Chan, K C, Fung, H, & Thapa, S (2007). China financial research: A review and synthesis. International Review of Economics and Finance, 16, 416-428.

Cheng, W Y, Cheung, Y L, & Yung, H M (2003). Profitability of the CRISMA system: From world indices to the Hong Kong stock market. Applied Financial Economics, 10, 45-57.

Cootner, P H (1962). Stock prices: random vs systematic changes. Industrial Management Review, 3, 24-45.

Corrado, C J, & Lee, S H (1992). Filter rule tests of the economic significance of serial dependence in daily stock returns. Journal of Financial Research, 15, 369-87.

Fama, E F (1970). Efficient capital markets: A review of theory and empirical work. Journal of Finance, 25, 383-417.

Goodacre, A, Bosher, J, & Dove, A (1999). Testing the CRISMA trading system: evidence from the UK market. Applied Financial Economics, 9, 455-468.

Goodacre, A & Kohn-Speyer, T (2001). CRISMA revisited. Applied Financial Economics, 11, 221-230.

Groenewold, N, Tang, S H K, & Wu, Y (2003). The efficiency of the Chinese stock market and the role of the banks. Journal of Asian economics, 14, pp 593-609.

Hiemstra, C, & Jones, J D (1994). Testing for linear and nonlinear Granger causality in the stock price-volume relation. Journal of Finance, 49, 1639-1664.

Hinkel, E, Hudson, R, Dempsey, M, & Keasey, K (1996). A note on the weak form efficiency of capital markets: the application of simple technical trading rules to UK stock prices - 1935 to1994. Journal of Banking and Finance, 20, 1121-32.

Huang, Y S (1995). The trading performance of filter rules on the Taiwan stock exchange. Applied Financial Economics, 5, 391-95.

Karpoff, J M (1987). The relation between price changes and trading volume: a survey. Journal of Financial and Quantitative Analysis, 22, 109-126.

Levy, R A (1967). Relative strength as a criterion for investment selection. Journal of Finance, 22, 595-610.

Lo, A W, Mamaysky, H, & Wang, J (2000). Foundations of technical analysis: Computational algorithms, statistical inference, and empirical implementation. Journal of Finance, 55, 1705-1765.

Lock, D B (2007). The China A shares follow random walk but the B shares do not. Economics Bulletin, 7, pp. 1-12.

Ma, S, & Barnes, M (2001). Are China's stock markets really weak-form efficient? Published in: The efficiency of China's stock market (2004), Ashgate Publishing.

O'Shaughnessy, J P (2005). What works on Wall Street: A guide to the best-performing investment strategies of all time. McGraw-Hill.

Pruitt, S W, & White, R E (1988). The CRISMA trading system: Who says technical analysis can't beat the market? Journal of Portfolio Management, 14, 55-58.

Pruitt, S W, Tse, K, & White, R E (1992). The CRISMA trading system: The next five years. Journal of Portfolio Management, 18, 22-25.

Raj, M, & Thurston, D (1996). Effectiveness of simple technical trading rules in the Hong Kong futures markets. Applied Economics Letters, 3, 33-36.

Rogalski, R (1978). The dependence of prices and volume. Review of Economics and Statistics, 36, 268-274.

Saatcioglu, K, & Starks, L T (1998). The stock price-volume relationship in emerging stock markets: the case of Latin America. International Journal of Forecasting, 14, 215-225.

Silvapulle, P, & Choi, J S (1997). Testing for Linear and Nonlinear Granger Causality in the Stock Price-Volume Relation: Korean Evidence. Quarterly Review of Economics and Finance, 39, 59-76.

Sweeney, R J (1988). Some new filter rule tests: methods and results. Journal of Financial and Quantitative Analysis, 23, 285-300.

Van Horne, J C & Parker, G G C (1967). The random walk theory: an empirical test. Financial Analysts Journal, 23, 87-92.

Van Horne, J C & Parker, G G C (1968). Technical trading rules: a comment. Financial Analysts Journal, 24,128-32.

Wong, M C S (1995). Market reactions to several popular trend chasing technical signals. Applied Economics Letters, 2, 449-456.

Wong, W K, Manzur, M & Chew, B K (2003). How rewarding is technical analysis? Evidence from Singapore stock market. Applied Financial Economics, 13, 543-551.