Two ADX Variations

The average directional index is a widely, used measure of trendiness. A test of two small variations of an ADX system showed that the absolute value of the ADX has no particular significance. The test used a $1,500 stop and allowed $100 for slippage and commissions on the following markets: British pound, coffee, cotton, crude oil, deutsche mark, Euro­dollar, gold, Japanese yen, U.S. bond, and wheat. All available data from January 1, 1975, through July 10, 1995, were used.

The first system tested entered long or short when the 14-day ADX value crossed over 30. A crossover occurs when yesterday's value is less than 30 and today's value is more than 30 for the ADX. If the 3-day SMA of the close is greater than the 20-day exponential moving average, then the system buys on the next day's open. Conversely, if the 3-day SMA. of the close is less than 20-day exponential moving average, then the system sells on the next day's open. The initial stop is at $1,500 and there is a $100 allowance for slippage and commissions. The exit signal was a trailing stop at the highest high or lowest low of the past 20 days.

The second variation does not rely on the absolute value of the ADX; it requires only that the ADX be rising. The ADX was assumed to be rising if it was higher than its value 28 days ago. Otherwise, the rules of the two variations are identical. The results are summarized in Tables

166 Developing Trading System Variations Table 5.11 Results for ADX > 30 trading system

Market	Profit (S)	Number of Trades	Percent­age of Wins	Win/ Loss Ratio	Average Trade (S)	Maximum Intraday Draw­down ($)	Profit Factor
British	66,188			3.72	1,298	-8,869	2.61
pound
Coffee	112,531			5.76	1,758	-13,065	2.81
Cotton	68,795			4.09	1,376	-8,355	4.09
Crude oil	17,780			2.93		-3,550	2.93
Deutsche	21,338			2.00		-7,238	1.56
mark
Eurodollar	18,675			2.13		-3,350	2.38
Gold	-1,860			1.85	-37	-25,180	0.95
Japanese	42,125			2.56		-9,887	2.25
yen
U.S. bond	2,525			2.05	-45	-14,588	1.05
Wheat	5,675			1.65		-6,950	1.28
Total	355,632
Average	43,745			2.87		-10,103	2.19

5.11 and 5.12. The rising ADX system seems to be more profitable than the ADX > 30 system. A t-test determined that the difference was indeed statistically significant, since t-statistic was 2.76, greater than the t-criti-cal value of 2.26.

These results show that the trend of the ADX is more useful than the absolute value of the ADX. However, using a value such as ADX > 30 is a restrictive condition that reduces the number of trades signifi­cantly. You will find that the ADX > 30 system is in the market for only 40 to 45 percent of the time spent by the rising ADX system. Therefore, the losses and drawdowns are smaller for the ADX > 30 system. Thus, both approaches have their strengths and weaknesses, and your choice will reflect your trading preferences.

Figure 5.9 and Figure 5.10, page 168, will help you visualize how the two systems work. Figure 5.9 shows the entry due to an ADX rising system in the November 1995 orange juice contract. The ADX indicator along with the reference level of 30 appear in the lower half of the fig­ure. It may appear that the ADX has fumed down, but it is rising on a longer time frame of 28 days. Figure 5.10 shows the entry after ADX has crossed over the 30 level. The entry signal from the ADX>30 system

Two ADX Variations 167

Table 5.12 Results for ADX rising trading system

Market	Profit ($)	Number of Trades	Percent^ age of Wins	Win/ Loss Ratio	Average Trade (S)	Maximum Intraday Draw­down (S)	Profit Factor
British	131,469			3.43		-24,850	1.85
pound
Coffee	217,465			5.28	1,021	-19,824	2.08
Cotton	57,495			2.90		-17,745	1.54
Crude oil	-1,590			2.05	-13	-14,620	0.97
Deutsche	74,888			2.07		-10,475	1.99
mark
Eurodollar	38,600			3.45		^t.000	2.57
Gold	33,150			2.46		-24,120	1.33
Japanese	71,850			2.94		-16,763	1.65
yen
U.S. bond	60,163			2.52		-12,188	1.49
Wheat	5,105			1.98		-24,131	1.04
Total	690,185
Average	85,635			2.91		-16,872	1.65

Figure 5.9 The entry point derived from a rising ADX entry system for the No­vember 1995 orange juice contract.

168 Developing Trading System Variations

Figure 5.10 Note how the ADX > 30 entry occurs almost 3 months after the rising ADX entry in Figure 5.9.

came 3 months after the signal from the ADX rising system. You should expect such delayed entries because the ADX is a heavily smoothed indi­cator. Clearly, trading styles and preferences will govern which entry method you prefer.

The ADX logic assumes you are trying to confirm the presence of a strong trend. These entries could occur after prices have made a small move, since the heavily smoothed ADX takes several days to react to changing prices. Many traders do not like to buy on strength or sell on weakness. These traders prefer to lower risk at entry by buying a pull-back or selling a small bounce in prices. In the next section we review a variation that tries to trade the "pullback" in prices.