The anatomy of returns from moving average trading rules in the Russian stock market

ABSTRACT

This paper examines the profitability of index trading strategies that are based on dual moving average crossover (DMAC) rules in the Russian stock market over the 2003–2012 period. It contributes to the existing technical analysis (TA) literature by comparing for the first time in emerging markets the relative performance of individual stocks’ trading portfolios with that of trading strategies for the index that consists of the same stocks (i.e., the most liquid stocks of the Moscow Exchange). The results show that the best trading strategies of the in-sample period can outperform buy-and-hold strategy during the subsequent out-of-sample period, although with low statistical significance. In addition, we document the benefits of using DMAC combinations that are much longer than those employed in previous TA literature. Moreover, the decomposition of the full-sample-period performance into separate bull- and bear-period performances shows that the outperformance of the best past index trading strategies over is mostly attributable to the fact that they managed to stay mostly out of the stock market during a dramatic crash caused by the global financial crisis.

KEYWORDS:

• Technical analysis
• moving average
• ordered weighted average
• trading rules
• index trading
• market efficiency

JEL CLASSIFICATION:

• G11
• G14
• G17

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Beside the BRIC countries, the sample data of Aye et al. (2014) also includes the South African stock market.

² Many different weighting schemes have been used for exponential weighting in TA literature (e.g., see Gardner 2006). In this paper, we use a widely-used variant as follows: w_i = 2/(i + n), where i refers to the serial number of the freshness indicator of the price quote within the timespan over which a MA is calculated and n is the length of the particular MA. We use this simple weighting scheme as it fairly takes account of the wide range of MA lengths employed in this study.

³ In contrast to EWMAs, which reduce the weights of observations exponentially when moving back through time, the weights of the observations for OWMAs are determined on the basis of their order of magnitude (see Luukka et al. forthcoming for details).

⁴ Because of the emerging status of the Russian stock market, we do not include the use of short positions in our analyses for two main reasons: During a part of the sample period, the Russian regulators banned short sales (see Kudrov et al. 2012, for details). In addition, even if short sales had been allowed, their transaction costs would most probably have been high enough to nullify their benefits, particularly during the out-of-sample period when the short selling ban was set. We tested this by assuming that short selling would have been permissible throughout the out-of-sample period using the average of short selling costs of 18 Russian brokers documented in Kudrov et al. (2012) as an estimate of short selling costs. The results showed that after the inclusion of transaction costs, the strategies also allowing short selling would have been outperformed by the long-only strategies even if short selling had been permissible throughout the out-of-sample period, which was not the case.

⁵ To eliminate ‘whiplash’ signals in cases when shorter and longer MAs are close to each other, we test all DMAC rules with a 1% band, which requires that the shorter MA must exceed (fall below) the longer MA by 1% before a buy (sell) signal is implemented (we also repeated the tests without a band filter. Generally, the results did not change much, and due to space limitations, we report only the results for the DMAC rules with a 1% filter).

⁶ The adjustment for skewness and kurtosis is made by multiplying the standard deviation in the denominator of the standard Sharpe ratio by the ratio Z_CF/Z_c, where Z_c is the critical value of the probability based on the standard normal distribution (set to – 1.96 to correspond to the 95% probability level) and Z_CF is the corresponding skewness- and kurtosis-adjusted value calculated on the basis of the fourth-order Cornish–Fisher expansion as follows:

where S refers to Fisher’s skewness and K to excess kurtosis (see Pätäri 2011, for the introduction of the SKASR and the related risk metrics skewness- and kurtosis-adjusted deviation (SKAD).

⁷ All the performance metrics are calculated on the basis of monthly returns in order to avoid some nondesirable characteristics of daily return distributions (e.g., high kurtosis and higher autocorrelation).

⁸ The percentage is based on the updated estimates of explicit (i.e., commissions and other fees) and implicit (represented for the most part by the price impact of the trades) trading costs for emerging market countries (documented in Domowitz, Glen and Madhavan 2001; Chong, Cheng and Wong 2010), allowing for the fact that the proportion of implicit trading costs to total transaction costs is remarkably lower in the case of index trading than in the case of trading on individual stocks. Given that the brokerage fees of Russian exchange traded funds are remarkably lower than the trading cost estimate employed in this study and that the transaction costs are generally clearly lower for institutional investors than for retail traders, the estimate employed in our study is on the safe side in the sense that the true after-costs returns would be downward rather than upward biased.

⁹ Israelsen (2005) suggests to multiply the excess return by its volatility in cases when negative excess return is negative in order to maintain the interpretation of the Sharpe ratios consistent throughout the distribution of the ratios (see the original article for details).

¹⁰ See the original article for details (Ledoit and Wolf 2008. The corresponding programming code is freely available at: http://www.econ.uzh.ch/faculty/wolf/publications.html).

¹¹ In this particular case, a sell (buy) signal is considered correct if the underlying index return has been negative (positive) when the signal has been ‘on’.

¹² Although the outperformance of SMA(132,199) over OWMA(113,200) is not statistically significant, the former dominates the latter in the mean-variance framework (i.e., the return of SMA(132,199) is higher, whereas its volatility is lower). The dominance also holds in the mean-SKAD framework. In addition to the better timing of the buy signal following the financial crisis period, the better performance of SMA(132,199) is also explained by the smaller average losses stemming from false signals (the number of which is equal to both of these two trading rules during this period).

¹³ The last sub-period from 24 May 2012 to the end of December 2012 is classified as bullish despite the cumulative return of the MICEX index from the previous trough not exceeding 20% by the end of 2012. However, the rising trend continued until January 2013, when the cumulative return of 20% from the previous trough was exceeded before the next declining trend.