The Pullback System

An old market maxim says, "Buy on a pullback into support." Many traders like to trade the pullback because it often provides an entry point with relatively low risk. Contrast this philosophy with trading the ADX systems, which buy on strength or sell on weakness. The pullback max­im, however, is not a precise statement of a trading system. We have to define what we mean by pullback and what represents support. This sys­tem is a variation of the CB-PB system examined in chapter 4. Here the

The Pullback System 169

pullback is defined relative to a moving average rather than after a 20-day breakout.

A pullback is simply a minor correction within an uptrend. The pullback itself could take many forms. For example, you can define pull-back as thrte consecutive down days. Perhaps you can define pullback as a "return to support" by a moving average. You can pick a variety of av­erages, such as a 20-day or 50-day simple or exponential moving aver­age. The term "return to support" is vague—you must decide if prices must touch the average, go below the average, or get within 1 percent of the average. Once you agree what "pullback" and "support" mean, you must then decide at what point to place your buy order. For example, you could buy at the next day's open, the next day's high, or at the 5-day high. Picking a precise definition will allow you to build many variations of this system.

We want to build a system that recognizes a retracement in both downtrends and uptrends. We will define a pullback as a new 5-day low in an uptrend or a new 5-day high in a downtrend.

Next, we must define the trend. We will assume that if the trend is up, the low remains above the d-day SMA when the market makes a new 5-day low. Similarly, the high will remain below the d-day average when the market makes a new 5-day high. We arbitrarily pick a 50-day SMA as the reference moving average because the prices will stay above or be­low this average during strong trends. Thus, the trend is defined as up when the market makes a new 5-day low but the low remains above the 50-day SMA (see Figure 5.11). The matching condition for the down­trend is that the new 5-day high remains below the 50-day SMA.

You can imagine other variations of this system. For example, you can add a trend filter, using the ADX as suggested by Connors and Raschke (see bibliography for reference). They suggest using a 14-day ADX value greater than 30 and using a 20-day exponential moving aver­age. Another option is to use a 14-day RSI or stochastic oscillator and to look for reversals. For example, you could look for the stochastic to fall below 20 and then rise above 20 to define the buying point.

The results of tests of the pullback systems using the same data sets as for all other systems in this chapter are shown in Table 5.13, page 171. We again used a $1,500 initial stop and allowed $100 for slippage and commissions. The entry was on the open after a new 5-day low above the 50-day SMA and vice versa. The exit was on the trailing 20-day high or low. The basic idea of the pullback system seems valid when you con­sider its profitability over a wide range of markets. The 20-year test pe­riod also confirms that the pullback system will find profitable trades in the future.

170 Developing Trading System Variations

-115*20 -112*16 -109*12 -106*08 -103*04 -100*00 -96*28

Figure 5.11 The pullback system had many entry points into the strong up­trend in the June 1995 U.S. bond market. The solid line is the trailing 20-day low. The 50-day SMA is shown by crosses.

The percentage of profitable trades is relatively low, about 20 to 30 percent. However, the low win ratio is offset by the high average win-loss ratio. The drawdowns are moderate compared to other systems we have studied. The pullback system seems to work best in markets that have regular corrections within choppy trends. The pullback system also likes markets that make swing moves. Note, for example, the small loss in crude oil and the small profit in the U.S. bond market. The other trend-following systems discussed in this book have tested poorly on these markets, which often have choppy trends.

In keeping with its design philosophy, the maximum drawdown fig­ure is substantially smaller than in other trend-following systems. How­ever, the results in Table 5.13 are not as attractive as some of the trend-following systems we have seen. The reason is obvious in the live hogs chart shown in Figure 5.12. The market did not make profitable swing moves after retreating to the 50-day average. Your exit strategy is more important in such markets that bounce weakly off the 50-day average. You can experiment with your favorite exits, or use discretion.

So far we have not taken advantage of a key design feature. Since we are buying or selling after a pullback, we hope to get a low-risk entry point. A low-risk entry point is ideal for using a variable-contract

The Pullback System 171

Table 5.13 Results for pullback system: new 5 day low above 50-day SMA or new 5<lay high below 50-day SMA; exit on trailing 20-day high or low; $1,500 initial stop, $100 S&cC

Market	Profit (S)	Number of Trades	Percent­age of Wins	Win/ Loss Ratio	Average Trade ($)	Maximum Intraday Draw­down ($)	Profit Factor
British	122,269			5.62		-16,569	2.21
pound
Coffee	205,035			9.34		-26,063	2.52
Cotton	46,630			3.96		-10,360	1.72
Crude oil	1,840			2.21		-12,070	1.05
Deutsche	30,725			2.82		-14,963	1.46
mark
Eurodollar	16,575			4.01		-4,200	1.66
Gold	34,320			4.04		-25,430	1.47
Japanese	27,575			3.03		-14,925	1.37
yen
U.S. bond	35,850			3.84		-15,400	1.43
Wheat	-7,588			2.70	-56	-21,131	0.85
Total	513,231
Average	57,869			4.16		-16,111	1.65

-48.00

-47.00

46.00

•45.00

-44,00

-43,00 -4200 -41.00

Figure 5.12 The choppy trend in live hogs triggered the system entries but did not generate significant profits. This is a common problem in choppy markets, and your exit strategy becomes even more important.

172 Developing Trading System Variations

money-management strategy. If you can risk $10,000 per position sig­naled by the pullback system, then trade up to a maximum of ten con­tracts using the 5-day high-low range as the measure of volatility. In Ta­ble 5.14, we traded a new 5-day high or low beyond the 50-day average. The exit was placed at the 20-day trailing high or low, and we used a $1,500 initial stop, and charged $100 for slippage and commissions.

There is a striking difference in performance among a one-contract strategy and a variable-contract strategy with a ten-contract limit. The profits of the latter strategy were more than six times greater. However, the drawdown increased almost eight times on average, so the gains come at a price. Note that the drawdowns are still smaller than always trading ten contracts. Thus, the variable-contracts strategy could be bet­ter than trading the equivalent number of fixed contracts. You can test other variable-contract approaches with this entry strategy to match your preferences.

In summary, the basic idea behind the pullback system is valid. It has relatively high win-loss ratio and average trade. yovl could try to

Table 5.14 Results for pullback system with a maximum of ten contracts

Market	Profit (S)	Number of Trades	Percent­age of Wins	Win/ Loss Ratio	Average Trade (S)	Maximum Intraday Draw­down ($)	Profit Factor
British	871,719			5.89	6,919	-67,419	2.74
pound
Coffee	794,998			6.68	5,039	-174,653	2.11
Cotton	406,250			4.63	3,031	-100,755	1.83
Crude oil	33,360			3.54		-68,100	1.13
Deutsche	297,975			3.79	2,207	-126,825	1.54
mark
Eurodollar	237,850			3.79	3,498	-35,750	2.34
Gold	129,290			3.99		-226,310	1.24
Japanese	86,288			3.26		-1 35,263	1.10
yen
U.S. bond	316,050			4.82	2,258	-133,200	1.65
Wheat	19,419			3.22		-162,260	1.06
Total	3,193,199
Average	352,642			4.36	2,511	-123,054	1.74

The Long Bomb — A Pattern-based System 173

add filters to improve system performance. You should consider a vari­able-contracts money-management approach to exploit the low-risk entry points.