The Long Bomb — A Pattern-based System

Many a football team has lived, and died, by the "long bomb." In Ameri­can football, this is a pass play designed to gain big yardage by going deep down the playing field. A successful long bomb can rekindle enthusiasm in your team, demoralize the opposition, and even get a quick score when little time remains. It is a gutsy play, but not without some risk. This sec­tion discusses a pattern-based system inspired by the philosophy of the long bomb. Pattern-based systems allow you to react immediately. How­ever, the patterns do not repeat frequently, and the market does not have to follow through on the pattern each time. You must also be careful when you test the pattern to use a sufficiently large data set.

The long bomb is a 5-bar, bottom-picking pattern. It is related to market action near double bottoms. The usual sequence is that a market probes for a bottom. As big players sense a bottom, they buy in "bulk" and produce a sharp rally. Other traders feel this may be an opportunity to short, and the smart-money-led rally is thwarted by short-sellers late to the party. As a result, the market retests the prior bottom. If there is no significant selling on the retest, the market is ready to resume an upward move, and does so with a strong surge. The strategy here is to get in during the upsurge—a simple enough idea. As it bursts out of the base, the market closes above the highs made over the past few days. If the following conditions are satisfied, buy tomorrow just above to­day's high:

1. Today's close is greater than the high of 2 days ago

2. Yesterday's close is greater than the high of 3 days ago

3. The close 2 days ago is greater than the high of 4 days ago

This relationship is perfectly general with no market-specific as­sumptions. In addition, there is no need to optimize the system. The fol­lowing twist makes this system interesting. Suppose you buy multiple contracts, betting $10,000 per trade. You can use the difference between today's close and the 5-day low, convert this "range" into dollars, and

174 Developing Trading System Variations

calculate the number of contracts. Thus, if the "range" is $2,000, you will buy five contracts and use a $10,000 hard-stop on the position. Here is another twist. You will exit at the trailing 50-day low to get a long-term system.

By now you should understand why the name of the system is the long bomb. By buying multiple contracts and using a very slow exit, we are following the TOPS COLA strategy, and, in football parlance, going deep.

The Power Editor code for Omega Research's Trade Station™ soft­ware is:

input: dx (50);

Vars: Numc(O);

if C - lowestClow.5) 0 0 then Num C = intportion(10000/((C-1owest(1ow.5))*Big

Pointvalue)) else NumC ⁼ 1;

Condltloni - c > h[2];

Condition? - c[l] > h[3];

Conditions = c[2] > h[4];

If Condltioni and Condition2 and conditions then buy

("12S Signal") NumC contracts at high + 1 point stop;

ExitLong at Lowest(1ow,dx)[l] - 1 point stop;

An input, dx, which is the number of days in the exit, allows us to experiment with the exit strategy. Just before we calculate the number of contracts, we should double check that today's close is above the 5-day low to avoid division by zero. The three conditions specifying the pat­tern are entered as inequalities. If the entry conditions are satisfied to­day, we will enter tomorrow just above today's high on a buy stop. Hence, we could see some slippage if the market opens beyond our stop, but this barrier entry condition acts as a filter and reduces the number of trades.

Figure 5.13 shows how this entry condition looks on a real bar chart, using the March 1995 Eurodollar contract just after a significant market low. The long-bomb pattern went long just after the retest of the low, after the first bounce. Notice that there was no retracement after entry: like the long bomb, it was up, up, and away. A similar pattern is shown in Figure 5.14 for the December 1995 heating oil contract.

Strange as it may seem, you will find that the long-bomb pattern oc­curs frequently at market bottoms. Table 5.15, page 176, gives the results of long-term testing over 10 markets. The period chosen is from January

The Long Bomb — A Pattern-based System 175

Figure 5.13 The long-bomb pattern enters as the Eurodollar accelerates out of its major bottom.

-53.00

52.00

-51.00

-50.00 -49.00

Oct

Nov

Figure 5.14 The long-bomb pattern in the December 1995 heating oil con­tract. The entry point saw good followthrough with little retracement.

176 Developing Trading System Variations Table 5.15 Long-term results for long-bomb trading system

			Number of
			Wins; Total	Maximum	Maximum
Market	Profit ($)	Profit Factor (Gross Profit/ Gross Loss)	Trades; Percentage Profitable	Number of Contracts	Intraday Drawdown (S)
British pound	114,313	1.54	12;35;34		-96,407
Coffee	417,770	3.03	9;34;26		-49,660
Cotton	222,590	1.98	10;31;32		-136,040
Deutsche mark	234,263	2.17	10;31;32		-60,725
Eurodollar	142,825	1.76	12;29;41		-72,200
Heating oil	25,544	1.07	12;45;27		-135,565
Japanese yen	779,324	5.93	11;24;46		-38,000
Soybeans	62,018	1.20	11;42;26		-120,695
Swiss franc	256,453	2.70	10;26;38		-42,310
U.S. bond	272,363	2.41	12;33;36		-63,400

1, 1982, through July 10. 1995, a 14-year period that spans a variety of market conditions. The results would be even better if we included a longer test period. We used a $1,500 stop, allowed $100 for slippage and commissions, and exited at the trailing 50-day low. We also calculated multiple contracts for a $10,000 investment, as explained above.

Table 5.15 does not show the fact that the average win/loss ratio was well above 3 for most of the markets. The results show that this sys­tem was profitable with multiple contracts, with considerable potential for drawdown. You can test a variety of exit strategies with this entry. You could also derive and test other patterns, such as a 3-bar pattern or 20-bar pattern.

Note here that we have around 30 trades in each market. If we as­sume that all markets are equivalent, we have 330 trades in Table 5.15, giving us reasonable confidence about system performance. Observe that these 330 trades occurred over a period of 14 years, averaging 24 trades a year, perhaps two a month. This may be a relatively low frequency for most trading programs. Hence, you should strive to develop a large portfolio of patterns, so that you will generate perhaps ten or more trades a month.

Testing patterns can be interesting, but must be done carefully. For example, there are a large number of candlestick patterns with tan­talizing names and exciting descriptions. However, they often do not withstand thorough testing. The long-bomb pattern shows that you do not need complexity to achieve your means—just a particular market intuition.

Summary 177