When do simple moving averages (SMA) serve as useful trading rules? Do they exploit some hidden pattern in asset price behavior? In their July 2011 paper entitled “The Trend is not Your Friend! Why Empirical Timing Success is Determined by the Underlying’s Price Characteristics and Market Efficiency is Irrelevant “, flagged by a subscriber, Peter Scholz and Ursula Walther investigate the relationship between the performance of technical trend-following rules and the characteristics (statistics) of the target asset return series. They use timing rules based on SMAs of different intervals (5, 10, 20, 38, 50, 100 and 200 trading days) as examples of trend-following rules. They consider the effects on SMA rule performance of variations in four asset price series statstics: the first-order trend (drift); return autocorrelation (return persistence); volatility of returns; and, volatility autocorrelation (volatility persistence/clustering). Analyses are long-only and ignore trading frictions, dividends, return on cash and buffering tactics such as stop-loss. They use a robust array of risk and performance measures to compare SMA rule performance to a buy-and-hold approach. Using both simulated price series and ten years of daily prices (2000-2009) for 35 country stock market indexes, they find that:
- SMA timing rules tend to beat a buy-and-hold approach when the target asset price series exhibits: negative drift (downward trend); high positive return autocorrelation (momentum); low return volatility; and, highly clustered return volatility. Volatility effects are weak compared to drift and autocorrelation effects.
- SMA timing rules impressively beat buy-and-hold when asset price drift is negative. (But why target an asset with expected negative drift?)
- Short-interval SMA trading rules benefit especially from strong positive return autocorrelation (momentum), but they suffer extremely if autocorrelation turns negative (reversal). The mean return autocorrelation in the 35 country stock markets during 2000-2008 is rather low, offering minimal enhancement through timing.
- Return skewness and kurtosis have no significant impact on SMA timing rule performance.
- There is no evidence that SMA timing rules predict target asset price series properties, only that they react to stochastic properties of the asset price series. The fitting of a timing rule to past price data thus seems tantamount to fighting a future war with strategies from past wars.
- SMA timing rules do not inherently reduce risk. While they tend to protect against extreme drawdowns in bear markets, they elevate other investor risks (mainly the risk of bleed-out via frequent small losing trades).
- During 2000-2009, SMA timing rules (coincidentally) work rather well in the developed world and quite poorly in emerging markets.
In summary, evidence from an array of tests on simulated and real stock market price series indicate that trading rules based on simple moving averages sometimes beat a buy-and-hold approach and sometimes do not, depending mostly on series drift and return autocorrelation and somewhat on return volatility. Such rules do not intrinsically enhance risk-adjusted performance.
Cautions regarding findings include:
- Results do not rule out the usefulness of SMA trading rules for investors who can somehow otherwise predict asset price series drift, return autocorrelation and volatility (for example, via analysis of company fundamentals or socioeconomic trends).
- Trading costs, dividends and return on cash may be material to long-term trend-following net performance.
