Does simple technical analysis based on moving averages work on high-frequency spot gold and silver trading? In their August 2015 paper entitled “Does Technical Analysis Beat the Market? – Evidence from High Frequency Trading in Gold and Silver”, Andrew Urquhart, Jonathan Batten, Brian Lucey, Frank McGroarty and Maurice Peat examine the profitability of 5-minute moving average technical analysis in the gold and silver spot markets. They consider simple moving average (SMA), exponential moving average (EMA) and weighted moving average (WMA) crossing rules. These rules buy (sell) when a fast moving average crosses above (below) a slow moving average. They start with four commonly used parameter settings, all using a fast moving average of one interval paired with a slow moving average of 50, 100, 150 or 200 intervals [(1-50), (1-100), (1-150) or (1-200)]. They then test all combinations of a fast moving average ranging from 1 to 49 intervals and a slow moving average ranging from 50 to 500 intervals, generating a total of 66,297 distinct rules. To compensate for data snooping bias, they specify in-sample and out-of-sample subperiods and test whether the most successful in-sample rules work out-of-sample. They also use bootstrapping as an additional robustness test. Using 5-minute spot gold and silver prices during January 2008 through mid-September 2014, they find that:
- Commonly used (1-50), (1-100), (1-150) or (1-200) rules generate negative returns for SMA, WMA and EMA when applied to 5-minute spot gold and silver prices.
- Among all 66,297 rules tested on 5-minute spot gold prices:
- 56.4%, 56.3% and 32.7% of the SMA, EMA and WMA rules, respectively, generate positive gross returns, but only 20.2%, 1.3% and 1.9% are significantly positive.
- The 44-332, 49-259 and 49-498 rules generate the highest gross returns for SMA, EMA and WMA, respectively.
- Among all 66,297 rules tested on 5-minute spot silver prices:
- None of the rules generate significantly positive gross returns.
- The 45-300, 48-380 and 49-347 rules generate the highest gross returns for SMA, EMA and WMA, respectively.
- An in-sample/out-of-sample test for spot gold, with break point at the end of August 2010, shows that the most successful in-sample rules generate positive gross returns out-of-sample. However, only the SMA rules are statistically significant during the in-sample and out-of-sample subperiods.
- An in-sample/out-of-sample test for spot silver, with break point at the end of September 2010, shows that the most successful in-sample rules are unprofitable out-of-sample.
- A bootstrap analysis generally confirms findings of other tests.
In summary, evidence indicates that it is possible (but not easy) to generate significant positive returns via high-frequency (5-minute) technical trading of spot gold, but not spot silver.
An ancillary finding is that SMA generally works better than EMA and WMA.
Cautions regarding findings include:
- The authors do not support the assertion that moving average parameter settings commonly used with daily data are also commonly used with intraday frequencies (in other words, that moving average parameters are fractal).
- Results are gross, not net. Accounting for trading frictions would reduce reported returns and may alter findings. In particular, short measurement intervals tend to generate more trades than long measurement intervals, so ignoring frictions favors short intervals.
- The analysis is purely empirical. The authors do not offer any theoretical bases for why some rules/parameter settings might work better than others.
- The study addresses only the statistical significance of profitability, not economic/benchmarked significance.
Frequently Asked Questions
What is a moving average crossover rule in technical trading?
A moving average crossover rule is a technical trading strategy that buys when a fast moving average crosses above a slow moving average and sells when it crosses below. The study tested simple (SMA), exponential (EMA) and weighted (WMA) versions on 5-minute spot gold and silver prices, using a fast average of 1 to 49 intervals and a slow average of 50 to 500 intervals.
Do moving average trading rules work on high-frequency gold and silver prices?
Moving average rules show some success on 5-minute spot gold but none on spot silver, based on January 2008 to mid-September 2014 data. For gold, 20.2% of SMA rules, 1.3% of EMA rules and 1.9% of WMA rules produced significantly positive gross returns, while no rule did so for silver. The authors conclude that profitable high-frequency technical trading of gold is possible but not easy.
How do researchers test thousands of moving average rules without data snooping bias?
Researchers in this study tested 66,297 moving average rules and addressed data snooping bias by splitting the data into in-sample and out-of-sample subperiods, then checking whether the best in-sample rules still worked out-of-sample. Break points were the end of August 2010 for gold and the end of September 2010 for silver. They also used bootstrapping as a robustness test, which generally confirmed the other findings.
Which works better for intraday trading: simple, exponential or weighted moving averages?
Simple moving average (SMA) rules generally worked better than exponential (EMA) and weighted (WMA) rules in the 5-minute gold and silver study. For gold, 20.2% of SMA rules were significantly positive versus 1.3% for EMA and 1.9% for WMA. Only the SMA rules were statistically significant in both the in-sample and out-of-sample gold subperiods.
What are the limitations of high-frequency moving average trading results for gold?
The reported moving average results for gold are gross returns, so trading frictions would reduce them and might change the findings, especially since short intervals generate many trades. The study addresses only statistical significance, not economic significance relative to a benchmark, and offers no theoretical reason why some parameter settings work better. The authors also do not support the assumption that parameters common for daily data are common at intraday frequencies.