| Home | My photos | Forex | My trading | Contacts |
|
|
Perry J. Kaufman. Smarter Trading. Improving Perfomance in Changing Markets | ||||
|
free download links about online stock trading, forex, futures, stock investing, market, trading systems Ask a trend follower, "How do you trade successfully?" The answer will be, "Let your profits run and cut your losses short." That seems to express the underlying philosophy behind trend-following programs. One of the most stunning examples is gold, rising from $100/oz to peak at $850/oz in January 1980. There were good opportunities to take prof its at $250/oz and $400/oz, but neither of these would have been close to the profits that were possible if you had simply held a long-term position using nearly any trend-following method. In Figure 5-1 (a), the same pattern can be seen in silver; and Figure 5-1 (b) shows an S&P move in October 1987 that is difficult to forget. Large profits represent what "might have been" in one price move. This chapter will show that waiting for the exception is not as good as PROFIT-TAKING Disadvantages • Miss big profits • More frequent profitable trades • Less slippage closing out trades • Less slippage than reversing a trade (e.g., long to short) • Better reward/risk ratio When deciding to use profit-taking, the only disadvantage is the fear of "missing the big move." Instead, trading can be improved. returning more frequent profits. Spectacular price moves do not occur often enough to be worth sacrificing a good, everyday plan. In addition, the great risk associated with a major move requires a constant invest ment that is too big to be reasonable for the few times it will serve to protect an infrequent occurrence. A Test for Profit-Taking This chapter will show that a simple rule for taking profits is better than waiting until the trend changes, which also means that profits will be taken before the "big" moves reach their peak. This will be done by testing a moving SPZ87.DI_Y-Daily
Figure 5-1. [Continued] Exceptional market moves, (b) The S&P drop of October 1987. Trend followers should have been short before the sharp drop, but the volatility of the following day would have made It difficult to justify holding a trend position. Looking back, it is clear that the trend was over on October 19. After that, a trend follower was just giving back profits waiting for a reversal. average system with profit-taking measured as a percentage of price. Other more complicated tests could look at different trading strategies as well as different profit-taking measurements, some based on volatility. First it is important to see that the idea of taking profits is sound. How It Was Tested A simple exponential moving average system was tested using Telerate's TeleTrac, and the results were compared and plotted using a spreadsheet. The formula for the trend was @exp_ma(price,period) = ema[1] + sc*(price - ema[1])
where @exp_ma is the exponential moving average function, saved as exp period is the number of days over which @exp_ma is calculated ema[1] is yesterday's exponential trend value price is today's closing price sc is the smoothing constant, sc = 2/(period +1) Six trend speeds were tested: 5, 10, 15, 25, 50, and 75 days. This dis tribution was selected to even out the percentage changes between the time periods in each test. The trading and profit-taking rules were: • Buy on the close when the exponential moving average turns up; sell • Close out the trade if the unrealized (open) profits, as of the closing • Do not reset a trade in the same direction once profits have been If the daily high and low prices had been used, profits could have been taken at the moment the profit was reached, instead of at the close. This could have improved results for many trades. But in this basic approach, all trades were exited at the close. If you use the interbank market, intraday data, or any 24-hour price stream, you will want to take profits as soon as they reach the target level. Here, we will only look at decisions made on the daily close of trading. Profit-Taking Levels To be certain that all levels of profit-taking were included in only a few tests, profit levels began at 100 percent of the trade entry price, then were divided by 2 until performance deteriorated. Because prices never doubled, no profits were taken at the 100 percent level, and those results will be used as a control to measure improvement. Profit-taking levels were: 100%, 50%, 25%, 12.5%, 6.0%, 3.0%, 1.5% For the serious technician, this method is also convenient for compar ing the frequency of profit-taking events with a normally distributed sample. Profit-Taking Measuring the Results
Although many traders judge success by the size of profits, these tests compare profits with risk. The goal is to find the greatest returns for the lowest risk. Portfolio managers would recognize this as "risk-adjusted" returns, discussed in Chapter 4. A better risk-adjusted return can be converted to higher profits simply by increasing the leverage, or reduc ing investment reserves, because of the lower relative risk. Expressing results in this way also allows investors to judge clearly how much risk they are willing to accept. In this example, normal commissions and slippage were applied to trades, but analysts must be careful to use realistic values for transaction costs to get correct results for their own trading. Profits, losses, and risk are shown in points. Risk is measured as one standard deviation of the daily accumulated profits and losses. The following three markets were tested: Hang Seng Index January 1988 February 1993
Deutsche Mark Index February 1988 February 1993 Unilever November 1987 November 1992 Coding the Tests The TeleTrac code is shown in Box 5-1 . This can also be done using a spreadsheet; however, it is important not to reenter a trade in the same direction once a profit is taken. You can solve this problem by keeping one spreadsheet column for the underlying trend direction (e.g., +1 for long and -1 for short), and another column to show whether that posi tion is active (+1 if holding the trade, 0 if closed out using profit-taking). It is tricky, but it can be done. Strategy testing programs, such as Omega's TradeStation or TeleTrac make it much easier. The Results The raw values (in points) of the test results are shown in Table 5-1 (a)-(c). The trend speed is the far left column and the profit-taking level is given along the top. The profit and risk in the "100%" column represent the trend system performance with no profit-taking. In each row of each of the three tables, there are tests that show increased profits; however, • item "DMX • date "DATA(date,880201,930226,item) • close "DATA(close,item) • speed 40 • exp Exp_ma(close,speed) • buy exp-exp[1]>0&exp[1]-exp[2]<0 • sell exp-exp[1]<0&exp[1]-exp[2]>0 • buysell Trade(buy,sell,sell,buy) • openbs Open_PL(buysell,close,0,0) • ptlevel 3.0 • ptrule openbs* 100/close>pt level • strategy Trade(buy,sell I ptrule,sell,buy I ptrule) • realized Clos_PL(strategy,close,0.001,0.001) • openpl Open_PL(strategy,close,0,0) • netpl realized + openpl • risk Std_dv(netpl,1200) • ratio netpl/risk Data used is the D-mark (indexed as a %)
Speed of the moving average Exponential MA calculation Buy signal rule Sell signal rule Trade strategy rules Open trade profit or loss Profit-taking level in % Profit-taking rule Final strategy rules Realized PL with trans costs Open trade PL on new strategy Net PL Risk = 1200 day standard dev Reward/risk ratio The code shows the calculation for the exponential moving average Using the "study" Exp_ma. The code first tests the simple trend change logic without profit-taking and calculates the open-trade profit, openbs. It compares the open-trade profits with the profit-taking level, then creates a new buy-sell rule named strategy. If the high and low prices are used to take profits on longs and shorts, lines 11 and 12 are replaced by the following three lines: ptlong (openbs[1]+high-close[1])*100/close>ptlevel ptshort (openbs[1 ]+close[1 ]-low)*100/close>ptlevel strategy Trade(buy,sell I ptlong,sell,buy I ptshort) these are also accompanied by increased risk. Table 5-2 presents the Hang Seng results as the return/risk ratio, making it easy to see whether the combination of risk and return gives better performance. Generalizing the Profit-Taking Patterns The values for the Hang Seng Index given in Table 5-1 are plotted in Figure 5-2. The slowest trend (75 days) showed best performance at the profit-taking level of 12.5 percent, the 50- and 25-day trends were best Table 5-1. Test Results
b. Deutsche Mark Results in Percent DEUTCHE MARK IMM INDEX, February 1,1988 , to February 26,1993 (in percent) MA 100% 50% 25% 12.5% 6% 3% 1.5% Days Profits Risk Profits Risk Profits Risk Profits Risk Profits Risk Profits Risk Profits Risk 5 9.83 4.09 9.83 4.09 9.83 4.09 10.47 4.04 5.79 4.63 10.88 4.98 10 8.79 6.34 8.79 6.34 8.79 6.34 9.96 6.57 11.00 7.36 10.15 7.58 15 23.47 8.04 23.47 8.04 23.47 8.04 26.73 8.59 26.58 9.49 25.90 9.90 25 37.76 12.01 37.76 12.01 37.76 12.01 37.92 11.91 30.93 9.97 22.79 8.68 50 39.61 12.75 39.61 12.75 39.61 12.75 50.24 14.49 26.00 8.00 21.59 7.35
5 240 1387 240 1387 240 1387 240 1387 4 330 1576 6350 1801 5620 1417 10 -1040 1313 -1040 1313 -1040 1313 -890 1313 300 1377 3530 1260 4960 1108 15 -3420 1630 -3420 1630 -3420 1630 -2440 1630 -2480 1645 1200 1214 1010 1232 25 -3360 1149 -3360 1149 -3360 1149 -2010 1149 -940 1107 6430 841 4640 676 50 -3620 967 -3620 967 -3620 967 -368 0 943 -3260 857 310 366 2940 431 75 -10060 1190 -10060 1190 -9710 1190 -11560 1092 -8080 1313 -2200 610 -2600 921 at 6 percent, the 15-day was best at 3 percent, and the faster speeds were still improving at 1.5 percent. Because slower trends allow larger prof its to accumulate, it makes sense that profit-taking levels will also be greater. Fast trends need comparably smaller targets. The consistency of the results is also reassuring. We can have more confidence in a method that begins with a logical premise and is con firmed by testing. Because risk declines faster than profits, the reward/risk ratio Improves. By using leverage, a better ratio can be turned into higher profits for lower risk.
Reasons for Profit-Taking The following are strong arguments for profit-taking: • Profit-taking has smaller slippage, because trades are closed out • It is not necessary to reverse a position from long or short, or short to • The more profit-taking, the less exposure to market uncertainty and • The extra profits that may be gained at the end of a move tend to have • The larger investment needed to wait for and then hold a longer, • Setting a smaller profit-taking level means that the system does not Closing out a trade with profits is not only personally satisfying but is sensible. As with other rules, it must be done properly. In a trend-fol lowing program, there are more losses than profits; therefore, the prof its must be larger than the losses; Liquidating a trade as soon as you see a profit will not result in a successful trading program. You must first test your system for a selection of profit-taking levels to see how the return/risk ratio changes. Then you have the basis for adding profit- taking to your plan. Profit Objectives versus Time The first part of this chapter showed that taking profits improves trend- following systems of all speeds. Longer trends develop larger profits; therefore they were helped by larger profit-taking objectives. Faster trends needed smaller goals. Using a small profit-taking objective in a long-term system causes small profits and relatively larger losses; a large profit objective in a fast trending system was rarely reached. The relationship between the time interval (the speed of the trend or the length of the trade) and the size of the profit-taking objective is sim ilar to the pattern of volatility over time (see Figure 5-3). As the period of evaluation becomes longer, the potential for profits levels off. This is the result of the maximum price move that would normally occur during that time period. Taking Profits at Only One Point It is easiest to think of taking profits at one price for each trade. This is often a fixed value or a percentage of the entry price. For example, buying
Figure 5-3. Size of profit-taking objectives for varying trend speeds. Profit-taking objectives Increase in size as the trend period Increases. This pat tern is very similar to risk or volatility over time. IBM shares at $60 using a medium-speed trend might have a profit-taking goal of $5 (about 8.3 percent). A foreign exchange trader might look for a profit of .0075 in the British pound, equal to .5 percent (one-half of 1 per cent). Using a percentage to specify the size of profits is a little more adapt able than using fixed values. Simple profit-taking goals are easy to test and simple to trade. The entire position goes on and off at the same time. But it is not as good as using multiple exit points, which give you more chances to succeed and reduce risk. Problems of Using One Prom-Taking Objective Single price objectives are impractical and often frustrating. If GM is bought at 30, with an objective of 33, what do you do if prices stop at 32 3 4, or at 32%? When are you close enough to your objective so that the few extra points are not important? A single profit-taking level is usually a number that worked "on average" over some historic period, whether it was a few weeks or many years. "On average" may have been very good in 1990 and very bad in 1994. Even clever profit-taking schemes, which adjust to volatil ity, may be out-of-phase for long time periods. Benefits of Using More Than One Profit Objective
The greatest benefit of profit-taking is the reduced risk gained by being out of the market. It follows that the sooner you begin to take profits, the faster your risk exposure will drop. To make this work, you need to trade a posi tion large enough to close it out in at least two parts, but preferably four parts. Two basic profit-taking strategies are shown in Figures 5-4 and 5-5. In both cases, a long position is taken at the same point—50.00—and a move of 2 percent is expected. Because price movement looks very similar to a random distribution, prices will move back and forth by 1 percent more than twice as often as they will reach 2 percent, and they will move 0.5 percent more than four times as often as they move 2 percent. This is important information when setting profit objectives. If, for every ten trades, you capture a profit of 2 percent only once, then you should have reached a 1 percent profit twice, and a profit of 0.5 percent four times. If the market was more volatile and you captured profits of 2 percent twice in ten trades, then you should have reached a 1 percent profit four times and a 0.5 percent profit eight times. It is important that we start by knowing how many times you reached the profit goal in your trading strategy. That gives us a way to know how often smaller profit objectives would be reached. This approach separates theory from reality. Look again at the number of times the profit levels would have been reached:
If a profit goal of 2% was reached twice in ten trades, then: Number Times Profit Total Reached Level Profits 2 2% 4% (known from testing) 4 1% 4% (2 extra) 8 0.5% 4% (4 extra) First, we see that the total profits remain the same. When you divide the profit objective by 2, you double the number of times you will reach that level. That is an important feature of a normal distribution. Of course, when profit-taking objectives are small and frequent, the transaction costs will be proportionately larger. That makes very small, frequent profit-taking unrealistic for real trading. Second, the four times that the 1 percent objective was reached include the two times that the price also reached the 2 percent level. Then there are only two additional times that prices reached the 1 percent level and did not reach the 2 percent level. Similarly, there were only four times that prices reached the 0.5 percent level without reaching the 1 percent level. Remember that the disadvantage is the added transaction costs due to more trades. Example. Two Profit Levels. Let us apply this to trading. Instead of taking a position with only 1 contract, 100 shares, or $1 million, take a posi tion of 2 contracts, 200 shares, or $2 million. Compare the result of using a single profit objective of 2 percent with two different objectives of 2 per cent and 1 percent, each for half of the position. At first we will look only at the profits that are taken: Case 1. Profit-Taking with One and Two Targets
From this, it would appear that setting more than one profit objective is the same as setting a single larger one, therefore, why not use the smallest ones? For two reasons: • You must still achieve an average profit of about the same size. Because the • The transaction costs become too high. The size of the profits can become As long as you can be profitable using smaller profit-taking levels, you have the following advantages: • The execution price is better. • More (unit) trades are profitable. For example, you will have taken some • Results are more consistent. Because market noise helps price reach • You don't have to worry as much about missing the profit-taking level by a • Overall risk is lower. Most important is that risk is reduced by being Adding Risk to the Picture We have discussed how the normal distribution of prices causes a clear pattern of how many times you can reach a preset target. Now let's look at the risk. It is safe to say that when prices move up 0.5 percent, they also move down 0.5 percent during the same time period. It would be especially cor rect if we did not have a very good timing strategy, or we selected an arbi trary period of price movement. In our 10-trade example, we assumed that, for each profit level reached, there was an equal risk. Looking at the trading profile for a trend-following system, there are 10 losses and only 2 profits: Case 2. No Profit-Taking
Note that the 10 trades each have a risk of 2 percent, because no profit- taking is used. In the next case, where profits are taken at two levels, the risk is always 2 percent unless profits are taken at a lower level, then the risk is equal to that level. For two profit-taking levels, the results are: Case 3. Two Profit-Taking Levels
Note that the bracket shows the 4 trades where profits were taken at the 1 percent level, and the remaining 6 trades where profits were not taken (where a full 2 percent loss is posted). This is the worst possible case, yet the return/risk ratio improved by 10 percent. This gain is the result of reducing the risk to 1 percent on a few units, rather than being exposed to a 2 percent risk on all trades. A 10 percent improvement in the reward/risk ratio is the same as adding 10 percent to your profits (with the same investment and same risk), or reducing the risk by 10 percent. With a Better Trading Strategy... What we have done is to look at profit-taking based on the number of profit levels reached in your current trading strategy. We have assumed that, when profits were not reached, prices were just as volatile in the wrong direction. A good trading strategy may only hold a trade for a short time. If profits are not reached within a preset number of days, the trade is ended. Therefore, if the market becomes less volatile and does not reach the 1 percent profit level, the trade can be closed out with a small prof it or loss—and never show a 2 percent risk. This would improve the reward/risk ratio significantly.
Stop-Losses A stop-loss that is smaller than the largest profit-taking level (e.g., 2 percent) can prevent you from reaching that profit level in 50 percent of the situations. Stops that are half the size of the largest profit level will cause the trade to be stopped out 75 percent of the time. This is simply based on the equal chance of a price move caused by market noise. Stop-losses must be used carefully, otherwise they interfere with performance. If your trading strategy is good, prices move in a profitable direction more often and do not reach the stop. If the entry timing is not as good, the stop-loss will be hit and prevent you from reaching your goal. More Profit-Taking If your trading program was able to take profits 4 times out of 10 trades, rather than the 2 times used in the example, the results of using profit-taking would be even better: Case 4. Using Profit-Taking with a Better System Performance
Profit-taking reduces risk, and more frequent profit-taking at different levels reduces risk more. In addition: • Transaction costs limit small profits. • Spreading profit levels over a broad range gives the most improve • Using a stop-loss that is smaller than the biggest profit-taking level • Limiting the time that you hold a trade reduces the trading risk by To find out how to get the profit-taking levels, you first need to test your own strategy assuming one profit point. That gives you the average of all profit objectives. You can then combine profit levels and position size to identify a simple pattern in the way profits should be taken. The most common plan is to have more targets at lower profit levels and spread them out further as profit targets become larger. The reduction in risk due to being out of the market may allow profit goals to be set much lower and still net very good returns with a better trading profile. |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| ©2007 Olesia | Home • My photos • Forex • My trading • Contacts |
