The design of a successful trading sjstem is based on a discoverj that yields a statistical edge. We find an edge through mmiber crunching, not from designing aroimd technical indicators. The objeaive is to find recurring patterns in daHy, weekly, or even intraday data. The process is iterative and painstaking, a ccgnitrve panning for gold.

The best sjstems alternate winning streaks with relatively flat periods of drawdown. Look for consistency across parameter seth and across time frames, and exploit as much historical data as possible. Experiment with combinations ofprofit targets and stop losses [30]. For example, ifthe profit taiget is 1.5 times the ATR and the stop loss is 1.0 times the ATR, determine the winning percentage and then chart the trade distribution like aprobability curve as shown in Figure 1.5. Plot the mmiber oftrades on the Y-axis and the percentage retum on the X-axis. Repeat this exercise for risk/reward ratios of 1:1, 1:2, and 1:3. The trade distribution plot should resemble a normal curve that is shifted to the right with apeakinprofitable trades to the right ofzero.

1 - t.4. l>i.iiil.tii

Ruies

After a sjstem has been designed, the entrj and exit rules must be defined. Each Acme trading sjstem conforms to a standard format with rules for both long and short positions, as shown in Table 1.6:

Table 1.6. System Rules

The professional trader experiences breatthrongh moments when all ofthe disparate technical elements that have been floating aroimd in ones mind foryears sjnthesize to produce inspired, original techniques. Some traders get there faster than others, but one day the trader realizes that money can be pulled out ofthe market consistently. Once the trader gets to that point, all ofthe external noise is eliminated. He or she stops going to chat rooms, turns offthe television, and cancels all sulBcriptions. Ultimately, the pursuit is just pure trading

As discussed. Average Tme Range (ATR) rs the standard for all Acme tradrng sjstems. Entrj and exit points are percentages of the ATR. Consequently, profit targets and stop losses must be adjusted to the appropriate time frame. For example, ifthe ATR of a stock is two points, then the profit taiget for a three-day holding period may be twice the ATR. The multiplier ofthe ATR for a profit target is a constant that is adjusted to the holding period.

Shnjlaiiy, the multiplier ofthe ATR for a stop loss is also a constant. For the day trader using a five-minute chart and a holding period of several hours, the profit multipUer may be 0.5 times the ATR and the stop multipUer may be 0.3 times the ATR. Day trading and sjstems trading are not mutually exclusive, as one might be led to beUeve.

Over time, the professional realizes that trading is a game of statistics and probability [12]. Traditionally, most trading books have focused on entrj and e:dt ixTints, e.g., a one-ixMnt stop loss or a 20o step loss When designing a trading sjstem, start with the goal of finding a strategj that is profitable 50° ofthe time, but the ratio ofthe average win to the average loss is 2:1.

The winning percentage changes as the risk/reward parameters are adjusted. In general, the lower the risk the lower the winning percentage wiU be. Tightening a stop reduces the mmiber ofwinners but reduces risk as well. In contrast, loosening a stop increases the mmiber ofwinners but increases risk. Trading is an equation-all ijarameters must be balanced to find the optimal siyp loss settings and profit targets. The trader must adjust the parameters to fit his or her risk profile. A higher winning percentage may feel more comfortable, but the trader may be sacrificing profit for comfort.

1 4 trade filters

The trader now has the decision of appljing trade filters to the sjstem. Depending upon the design ofthe sjstem, certain filters are more rele-ant than others For example, one ofthe swing trading sjstems. Acme N, is the only one using the ADX. Since N is apullback sjstem, the performance is directly proportional to the minimum price, historical volatDitj, and ADX. The higher these values are set, the better the results wiU be [4]. In contrast, the Acme R sjstem is based on the rectangle, a consolidation pattem where a higher ADX does not improve ov-eraU performance.

Table 1.7 shows the filters for each trading sjstem The Acme P sjstem is the only sjstem that does not use trade filters, but a volatihty filter could be apphed to it. Each sjstem was tested on ah of the trade filters-the ones that improv-ed testing results were kept, wldle the others were discarded; however, there may be other filters that could further impro\e performance.

Table 1.7. Acme Trade Filters

A m Average True Range

M-l Moving Average

/ Minimum Price

HI Historical Volatility

m Narrow Range

AI?X Average Directional Index

Directional Movement Index

The most interesting comparison of trade filters was the difference between the Moving Average filter and the Directional Movement Index filter. Some swing traders use the DMI to determine whether a stock is in an up trend or a downtrend. Overall, the performance of the moving average- filter (above or below the average) was better t n the DMI filter (positive or negative D I ratio) The

Acme N sjstem uses the mo\ing a-erage filter as an altemative to the DMI. A combination ofthe MA and DMI filters would further impro\e performance but reduce the number of signals.

The trade filters are grouped into two categories: price filters and technical filters. The ATR, MP, and NR filters are price filters derived from a stocks trading price and range for the current bar. The MA, HV, ADX, and DMI filters are technical filters based on historical price calculations The trader is free to modify the code to add other filters.

Note that the FiliersOn parameter is an input parameter to each ofthe Acme sjstems. By tuming this parameter on or off, the trader can compare the performance ofthe raw sjstem versus the filtered one.

Ayerage TnteRange {A TR)

The range of a bar is the difference between its high value and low value The True Range factors in any gap between the current bar and the previous bar If the current bars high is lower than the previous bars close, the n the ATR calculation uses the previous bars close as the True because ofthe gap down. If the current bars low is higher than the previous bars close, then a gap up has occurred, and the previous bars low is the True Low. Thus, the Tme Range is the difference between the Tme High and the Tme Low. FinaUy, the Average True Ran is the average ofthe Tme Range over a range of bars, e.g., twentj as shown in Figure 1.6.

vrsnlast4}a>ly 03/01/2002

40.000 38.000 2-01

-28.000 24.000 2d.o00 2.0500 2.5000

nil,.I

l-lb„..- l.h. Av,..... .,,,,.,

Average Tme Range is a measure of volatilit}. One might assume that a higher ATR imphes a more volatile stock, but wMle ATR is a good rrritial voIatiUt} screen, a better screen is to divide the ATR into the stock price. So, if Stock A has an ATR of two and a price of 50, and Stock has an ATR of two and a price of40, then StockAhas a VolatilHyPercentage ( ) of2 / 50 = 40o, and StockB hasaVPofZ 40 = 5°o. Consequently, StockB is more volatile.

Fortunately, even after stock prices com-erted from fictional to decimal in 2001, many stocks continue to have large daHy ranges. In 1995, the ATRs ofthe popular companies to trade (SunMicrosjstems, 3Com, and Applied Materials) ranged in the vicinity of three to four points. In 1999, many Nasdaq stocks had double-digit ATRs, some ofwhich are shown here:

- Redback Networks (RBAK:Nasdaq): 16

- Yahoo (YHOO:Nasdaq): II

- eBay(EBAY:Nasdaqj: II

- Copper Mountain (CMTN:Nasdaqj: 9

- CMGI. Inc. (CMGI:Nasdaq): 8

Only three years later, we stiU find it difficult to believe that stocks were having daHy ten-point swings. Since the heady dajs of 1999, the ATR ofthe tjpical momentum stock has declined to two or three points again as ofthis writing in early 2002.

MovingAverage (MA)

Much of technical analjsis is salf-futfiUing. The professional traders job is to watch what other traders are watching. Because the 50-day moving average (MA50)is so closely monitored, signals that occur here should be more profitable percentage-wise. The general principle is that a stock in an up trend tends to pullbackto theMASO as a support level (Figure 1.7). In contrast, a stock in a downtrend wiU pull up to theMASO as a resistance level (Figure 1.8).

The Acme F. N. and V sjstems use the 50-day moving average as a trade filter. The rules are simple. If trade filtenng is on, then along entrjis allowed ifthe stock is trading above its MA50. Similarly, a short entrj is allowed only if the stock is below its MA50. Because ofits importance, the moving average is a pattem qualifier for the Acme M System. It alerts the trader to a stock near its average by placing the letter "A" above and balow the bar.

Technicians use the 50-day MA to take positions on either side ofthe fine. If a stock in a long uptrend breaks down below the average, then a trader goes short. If a stock in a long downtrend breaks above the average, rhen the trader goes long. As with any strategy in the market, however, nothing is ever that

W.-I........ ,-...„....,,.,« m.l.<v,..........»..

simple. TheMASO gets penetrated often in either direction, and the prevailing long-temi trend usually wins out.

llast-daily 11 15«001 « 1 lfc«(CI03e,50,0) . ?

acnwlx

40.000 . 0 . 0

-lefstop- 7.000

1*"

34.000 33.000 32.000 31.000

Rgurel.7.LongEntrjat 50-day MovingAverage

The best way to determine whether or not the trend lias dianged is to use an ATR factor for confirmation. To confirm an uptrend, do not go long until the price exceeds one ATR above the average. Likewise, for a downtrend, do not go short until the price faUs one ATR below the average.

I.K.MKMI l-»m .11 Ml <U M..vM.)-.A