One popular method of using moving averages in forex trading is to plot crossovers for the generation of buy and sell signals. In this method, two moving averages with different time period inputs are plotted atop a currency pair’s chart. When the shorter-term moving average rises above the longer-term, a buy signal is generated; when it drops below, it generates a sell signal.

Although this may not be the most logical method of using this particular indicator (see Chuck LeBeau’s article, “Moving average crossovers may not be the best entry signals” for a thought-provoking analysis), it is a commonly used system, and in technical analysis, what often matters the most is not the logic of the method but the number of traders using it.

With this in mind, some recent experimentation with the AUS/USD daily chart and various moving averages was undertaken, with the intention of discovering which combination of time periods gave rise to the most profitable trades. Three datasets were tested: 10/30 days, 21/89 days, and 30/100 days.

In the charts used as illustrations, below, the shorter-term moving average is always the green line, and the longer-term is always the yellow, and therefore the buy signal is generated when the green line crosses above the yellow one.

As shown immediately above, the 30/100 moving averages generated buy signals, marked with vertical red lines, on 2 October 2007 at around 0.8850 (exit near the top of the trend at 0.9200) and on 15 February 2008 at 0.9025 (top near 0.9350), for a profit of 675 pips. This is not bad by any standards; however, the use of such long-term moving averages destroys much of the indicator’s sensitivity, generating buy signals more than halfway into the two strong uptrends on the chart and reducing the profit potential significantly.

The 21/89 moving averages generated buy signals on 28 September 2007 at 0.8780, and on 12 February 2008 at 0.8975. Using the same exit points as above, this gives a profit of 795 pips, rather more than the 30/100 dataset; however, greater sensitivity in the indicator could still generate greater profits.

The 10/30 moving averages generated buy signals on 13 September 2007 at around 0.8340 and on 4 January 2008 at 0.8775 (top near 0.9350), for a profit of 1435 pips, more than twice the amount of the 30/100 dataset. However, the second entry point, rather than being at the beginning of the uptrend, actually comes toward the end of the previous consolidation period, and if a forex trader had entered the market at that point, January would have been a difficult and disappointing month indeed.

Because moving averages are lagging indicators, showing where the price action has been in the recent past, they are not capable of calling precise tops and bottoms even at their most sensitive. They are best used in trending markets, rather than a range-bound one.

An important point in actual forex trading is that no indicator should be used in a vacuum, but rather as part of a body of evidence, built to establish the likelihood of profitability in any given trade situation. Each trader must determine his or her own level of risk acceptance, and level of sensitivity for the dataset when using moving averages.