That the trend is your friend is the first rule of trading. One of the easiest and most visual methods of trend following is to locate the moving average of the price data and trade with it. Experts have declared that trend trading is dead, or at least seriously injured, but one real-world experiment shows that with the right filters trading the trend as described by a set of moving averages is still profitable.
Consider the five-minute candlestick chart of the EUR/USD in “Noise reduction” (below). The blue line is the 10-period exponential moving average, the purple line is the 20-period exponential moving average and the red line is the 200-period exponential moving average.
Moving averages ultimately are useful because they are easy to follow: They smooth price action through a period, thus cutting out the price “noise.” Price noise is a term for excessive price volatility that may disguise a price trend.
The use of moving averages by traders is not new and many traders rely on moving averages as part of their trading tool kit. This article will define a simple trading plan, using the five-minute EUR/USD candlestick chart and the previously mentioned moving averages. In the style of “keep it simple, stupid,” we will demonstrate that this basic trading plan, if followed, will make money through time. The concept of the moving average in trading is not dead.
What are moving averages?
A moving average is an indicator that will calculate the average price of a commodity (in this case, the EUR/USD) throughout a period of time. To illustrate, because we are using the five-minute candlestick chart, the moving average is an equal weight of the past 10 periods of candlesticks, or the past 50 minutes. With each new candlestick, the oldest data point is dropped and the newest candlestick of data is added. Thus, a moving average is not static; it is rolling.
A simple moving average for M candlesticks of data shows the closing price of each candlestick (M1, M2,..., MD) is M, and where D is the total number of measurements made and MD is the most recently made measurement.
For our discussion, we are using exponential moving averages (EMA), the 10- and 20-period EMA, which are calculated similar to the simple moving average but give more weight to the more recent price action. The EMA is an attempt to reduce the lag of the simple moving averages: to make the moving average trendline respond more quickly to changing price action.
It’s helpful to transact with two moving averages, one of a shorter length than the other, to generate trading signals. This method should work well with trending commodities, and the EUR/USD and the other five major currencies are trending markets.
Our rules are simple: when the shorter of the two moving averages crosses over the longer of the two moving averages, a buy signal is generated. In the converse, a sell signal is generated. Because we are day trading the EUR/USD, we use an even shorter EMA against a short EMA: the 10-period against the 20-period EMA. The length of the moving average chosen should fit the time cycle traded.
When the 10-period EMA is above the 20-period EMA, we buy the EUR/USD (10 > 20 EMA). When the 20-period EMA is above the 10-period EMA, we sell the EUR/USD (20 > 10 EMA). Notice in “Noise reduction” how the signal of the trend change is definite and clearly observable. Further, the wider the spread between the two EMAs, the more likely the trend will continue.
John Murphy, author of the technical trading bible, Technical Analysis of the Futures Markets, linked cycles and moving averages: “There appears to be a definite relationship between moving averages and cycles. For example, the monthly cycle is one of the best known cycles operating throughout the commodity markets. A month has 20 to 21 trading days. Cycles tend to be related to their next longer or shorter cycles harmonically, or by a factor of two. That means that the next longer cycle is double the length of a cycle and the next shorter cycle is half its length.”
Thus, this trading plan uses the recognizable 10- and 20-period EMA. We are not bottom or top pickers; we only surf the trend, hopefully to take out the sweet spot. We want the trend to be easily recognizable to the trading community, so that, like lemmings all will follow it. Even with a clear buy or sell signal, the trader must know when to take profits. A simple trade plan would be to buy when the 10 > 20 EMA, and then when the 10-period EMA crosses to the downside (now the 20 > 10 EMA), the trader would sell the position and take profits. We also did the converse on the short sell.
We decided to backtest this strategy, because a simple eye-ball of the chart looked like it didn’t make sense. Viewing all five-minute candlestick chart data (high, low, open and close) from Nov. 28, 2001, until May 31, 2005, we found 5,736 trades on the upside and 5,735 trades on the downside. We paid the bid/ask spread to enter the trade. The returns are alarming:
Long: 10>20Short: 20>10
Number of Trades5,7365,735
Min P/L(133) tics(127) tics
Average P/L(3.06) tics(3.67) tics
Max P/L213 tics178 tics
STD P/L19.84 tics18.45 tics
The above data suggests that trend trading as a trading strategy doesn’t make sense for the EUR/USD. The average profit for the long or short trade was approximately three ticks, which is the bid/ask spread. Maximum losses were more than one cent, about 130 tics, which a day trader would never tolerate in a single trade.
In fact, with a nod toward our academic colleagues, the data without filters suggests a random walk. The standard deviation is about 20 to 19 ticks for the long and short trades; thus, it may make sense that the trader should use a firm stop and limit when entering the moving average trade. Even disregarding the data and reviewing the chart, we can see that the moving averages are lagging indicators and respond slowly when the market reverses. We need a filter for the simple trading plan that would allow us to protect profits and limit losses, and tell us during what time period to trade while following the trend.
Time is on our side
Certain time periods of the day are not well suited for trend-trading signals (see “Narrow opportunities,” below). From our research, we have found that trend trading normally in the late afternoon through mid-evening does not produce enough girth between the moving averages needed to obtain the space for a profitable trade.
The best trend trading times for the EUR/USD seem to be in the London market, usually around 2:30 a.m. EST time until 6 a.m., and then when the New York desks come in around 7 a.m. through about noon. Research suggests the two EMAs in question must have an approximate six-tick girth for the trade to be profitable. If visually the two EMAs are what might be casually described as “right on top of each other,” then there is no trade. The raw research presented earlier in this article does not filter for the girth of the moving averages and does not filter for time.
“Time-based trades” shows our trend trades just in the 8 a.m. to noon period. Two other filters should also be considered: limits and stops. Assume the trader shorted the EUR/USD on the crossover signal and got the price of 1.2068. The trick then is to know when to get out of the trades and take profits. Following the above trade suggested on the 20 > 10 EMA cross, the trader would then sell the EUR/USD. If the trader sold on the 20 >10 EMA cross and bought back when the 10 >20 cross, he would have made about 14 ticks. However, he could have made more than
20 ticks if he allowed the trade to run a little longer. We know that visually, the wider the girth of the moving average — the spread between the 20 EMA and the 10 EMA — the more likely the trade will be profitable.
We were able to test what the stops and limits should be using a five-minute double crossover method of moving average trading during our selected New York morning time frame. The summary of our results for buys only is located in “Profit matrix” (below). A buy signal is located when the 10 > 20 EMA.
For day traders, the most profitable performance came from a stop of 10 ticks and a limit of 50 ticks. However, swing trading proved the best use of the double crossover EMA: Stops of one cent or a cent-and-a-half and limits up to 100 ticks. We found this research to be confirmed on the one-hour chart. It should be noted, confirming common wisdom, that a deeper pocket when trading works better. Holding a trade with a wide stop loss takes deep pockets indeed. Research also suggests there is lots of noise within the trend; a stop of 20 ticks is not wide enough to plow through the noise to make the profit.
Trend trading is a lagging indicator method. Candlesticks denote entry and exit signals and should be used in conjunction with the above methodology and to compensate for the lag. It should be noted that we have not incorporated any reversal candlestick patterns into our analysis. In practice, we do incorporate these patterns. But in the effort of keeping it simple, we know that bulls win, bears win, but pigs lose. Every trader should locate the trend and grind out a result.
Final trading rules
Here are the rules we followed to generate our ultimate trading results.
• All data is from EUR/USD, five-minute chart, Nov. 28, 2001 to May 31, 2005, for the daily time period 8 a.m. to 12 p.m. (EST).
• Buy signal occurs when 10-period EMA crosses over 20-period EMA on five-minute period chart. Buy the next candlestick.
• Moving averages are calculated using typical price for each five-minute period. Typical Price = (1/3)*(Close)*(High)*(Low)
• Close signal occurs when user-define limit or stop-order hits:
Notes: A three-tick spread is included in calculations. For example, if you buy at 1.2010 with a stop loss of five ticks, your trade will hit stop and close at 1.2005. Therefore, the user-defined stop must be less than three ticks.
Open trades that are not closed by 12 p.m. (EST), are assumed to be “frozen” until 8 a.m. on the next trading day. If by 8 a.m. the market price exceeds the user-defined stop or limit, the trade will be closed immediately. If, during the frozen period, the market price exceeds the stop or limit and retraces back within the trade’s trading range (not violating stop and limit) by 8 a.m., the trade will remain open.
There is no consideration given to interest expenses that occur when trades are held overnight.
Prof. Leslie K. McNew teaches at the A.B. Freeman School of Business at Tulane University and runs the Pelican Fund.
Juan Londono, Parag Patel and Justin Tannen contributed to this article.