back start next


[start] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74] [75] [76] [77] [78] [79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94] [95] [96] [97] [98] [99] [100] [101] [102] [103] [104] [105] [106] [107] [108] [109] [110] [111] [112] [113] [114] [115] [116] [117] [118] [119] [120] [121] [122] [123] [124] [125] [126] [127] [128] [129] [130] [131] [132] [133] [134] [135] [136] [137] [138] [139] [140] [141] [142] [143] [144] [145] [146] [147] [148] [149] [150] [151] [152] [153] [154] [155] [156] [157] [158] [159] [160] [161] [162] [163] [164] [165] [166] [167] [168] [169] [170] [171] [172] [173] [174] [175] [176] [ 177 ] [178] [179] [180] [181] [182] [183] [184] [185] [186] [187] [188] [189] [190] [191] [192] [193] [194] [195] [196] [197] [198] [199] [200] [201] [202] [203] [204] [205]


177

sjstem stop to be very far away. Large stops may help reduce the risk of a price shock, but not ahvajs. Stops may reduce the performance volatihty of a portfolio simply by causing more time out of the market.

Theoretically, the pmpose of using a stop-loss order is to limit each loss to a controlled amount, and generate a performance profile with a reasonable ratio of average profits to average losses (for example, 2 : 1). In addition, a trader would like to have at least one-half of all trades profitable. Unfortunately, the market does not work this way. It is not possible to control both the reward/rid; ratio and the percentage of profitable trades.

Noise is the term given to erratic and unpredictable price movements. It is most often associated with relatively small moves, but can actually be any size. Noise is caused by traders entering and exiting the market with different objectives and different time horizons. For example, an institutional investor adds positions because of new funds from its clients; Russia sells gold to pay for a wheat purchase; or, an auto company liquidates part of its stock portfolio to cover foreigo exchange losses. Because each event is unrelated to the other, and there are so many of them, the total picture of this noise creates a random disfribution. This pattem has many of the properties of a random number disfribution, in particular, erratic prices (noise) that are twice as laige occur Iralf as often

Total Losses Are Constant

If noise is very similar to a random disfribution, then the number of occurrences times the size of the move will be very close to a constant value. The following is a tjpical result of setting stops of different sizes:

In reality, this pattem may varj slightly from market to market; however, the cost of slippage will usually be large enough to cause the net losses to be similar. In addition, the

Size of Stop No. 4f Occurrences Size toss

5 pips 20 100 pips + slippage

10 pips 10 100 pips + slippage

20 pips 5 100 pips + slippage

more frequent frades must also have additional transaction costs, with a chance of larger slippage from time to time; therefore, the sfrategj that uses smaller stop-losses should perform worse than the strategj with larger stops.

A Stop-Loss Con Conflict with the Trading Sfrategj

Combining a small stop-loss with a frend-following sjstem is very likely to conflict in piupose. if a trending sjstem gives one new frend signal each month, but the stop-loss is so small that it would be reached once each week, it would be very difficult to stay with a sjstem position. Once stoppedout, if the market turns and moves in the direction of the frend, the sjstem is proved right but you have no position. It is necessary to create a reentry strategj that can be unnecessarily complex

If every time you are stopped out of a trade, the price continues in the same direction until a reverse position is signaled, then it follows that the selection of the frend period is too slow. Rather than exit the trade when the stop is reached, the sjstem should be reversing its position. By shortening the frend calculation period, the need for a stop is eliminated.

Testing Stops over Ten Years

Table 23-3 gives the panem of performance for a simple moving average trading sjstem with calculation periods fran 5 to 300 days and stops from .02>oto 10"o. Four major maikets were selected, two stocks and two futures markets. The smallest stop-loss of 2>o appears on the left, and the largest of I0f6 and no stop NONE appear on the far right. Overall, the performance of the frend program without a stop is most consistent for ttie range of frend periods, although specific stops can be better for some intervals. It may be noted that there is a pattem in Siemens, in which the best performance runs diagonally from the upper left toward the bottom right, indicating that the size of the optimum stoploss may be related to the speed of the frend, but this pattem is not the same for the other markets tested and therefore, it may be the result of a single frade, such as the cradi of 1987, and not a robust relationship. Table 23-4 shows the same results for the Deutschemark and Eurodollars in terms of ridi-acjusted retums, in which ttie retums are divided by one standard deviation of the equity changes. In general, these results were the same as the unadjusted ones in Table 23-3.



Managing Risk with and without Stops

Despite the conclusion that the placement of fixed-level stops causes a trade to be stopped out arbitrarily, it is difficult to trade without a clear idea of rid;. For longer-term positions, a fixed stoploss may help in the case of catastrophic rid;. Before a stop-loss is used, it should be tested and compared with sjstem performance that does not use a stop. Careful selection of the trading strategj may provide adequate risk control with a separate stop-loss.

While fixed stops do not seem to enhance a trading program, a stop based on siport and residance levels may be quite different, it is alreadj known that a breakout sjstem, which enters on a new high and reverses on a new low, assumes the rid; of the range defined by the difference between the high and low. This is analogous to the size of a support-resistance range. Because this range varies with mari;et volatility and offers a view of risk that contrads with a trend sjstem, the combination of the two may improve net performance.

The safest way to reduce rid; is simply by reducing the size of your positions. Trading a smaller portfolio, a lower maigin to equity ratio, is the simplest approach to rid; reduction and does not conflict with the underljing trading strategj. in the final analjsis, the one who can profit by trading the smallest amount will be hurt the least when a price shock hits the mari;et.

TABLE23-3 Results of Tests Comparing Trend Period and Size of Fixed Stops* Annualized Rate of Renim (Adjusted for 25" Drawdown)

SufOoa in Whole Percent

J02 JOS .10 1 2i SO 100 LOO 4.00 7M 10.00 NONE

a.a<iyslerMot«r.2J29 Dayi (IKSI84bi 3116193)

S -26 -2.8 -2.7 -2,7 -27 -2,8 -2,8 -27 -2,8 -2S -2S -2S

10 -2S -2S -28 -28 -28 -28 -1.8 -28 -28 -28 -28 -2S

IS -2S 2S 2.3 2.7 J.a -2.8 -16 -2.4 -2.4 -2.4 -2.4 -2.4

SO -2.4 -2.4 -26 -2.4 -26 -2.4 -ZI -Z3 -2.2 -22 -11 -2.2

7S -13 -11 -1.1 -M -1.5 -IS -I* -1. -1 -1 -1.5

100 -1.5 -1.5 -1.5 -1 -1.7 -19 -.6 -7 -.5 -.6 -.6 -

300 2.2 2.2 2,1

IJ H IS

-2.5 -2.5 -2,S -25 -IS -2.5 -2.5

2,7 27

b.S™eih,l.«2D(iyi(2a;e7io 1 3192)

117 110 1l.<

25 101 101

94 6S 74

10-6 8,5 ei 7-5

5,2 4,8 49 4,1 4,9 49 5,0 5,0

4,8 59 59 5.3 3.0 47 7.e 77

rt:4S70 Day, (t 1/24/81D, 1413192)

-19 -19 -19 -19 -19 -1 -1- -1-8 -IS -1-8 -I* -14 -II -1 -1 -I.I -.7 -A - - - --6

52 69 M 6,4 17 1 11 ii.S 10,8 1 10,8

100 S,3 7.2 85

\A 8.1 6-8 7.7

:2.5 « { 0 12 4 )

3.8 .0 6,7 61 8,2 7.4

ID,0 9.9 M

200 73 5,4

29 59 6,5 6,5

5,4 5,4 5,4 5,4



TABLE 23-4 Comparison of 4-Year Results* Comparison of Cadi Retum versus Rid;-Adjusted Retum

i[0f>-L0!s Who Percent 02 X>5 10 15 .25 50 IM 2 0 400 7.O0 10.00 NONE D«m<J.»ni»fc ).03ll Day. (11121188 to 11 31

flnnuolued fiote of Relum (on Cash, m Percent)

10 -6,8 -S.0 -5.J

-8.3 -9.9 -S.5 -71 -7 1

-7.6 -7.0 -44

5.6 10.2 99 10.1

43 -4.3 -4.3

Anm/ofized ROfl (Adjusted for 25% DnwdownJ

5 -3 7 - -3 6 - -4.0 -3.6

10 -39 2.8 -29 -3.2

3.6 -36 -3-6

-3.5 -2.5 -2.4 -2.4 -24 -24

25 2.7 4.8 46 4-6 46 3-3 7 1 10.9 109 10.9 10.9

50 2 7 23.3 22 5 25.2 217 17.3 16.0 166 15.5 15.5 15 5

75 4.4 3.6 2.2 22 7 14.5 75 6.6 44 4.2 4.2

100 2.8 e.7 7.0 6.B IBI 9-5 IB.I 113 19 1 16.6 1B.6

ISO 37 24 1.7 17 -17 12.1 110 14.6 lO.S 83 8.0

200 14.0 22-0 193 19.2 17.1 134 132 15 1 9.1 59 5.3

250 147 IBS 17.6 167 160 12.8 99 129 74 8.2 72

1,042 Deyi (1422188 to I2IM¹)

Anwakzed Rate of Retum (on Cosh, Percent)

AnnuettftJ ROR (Adjusted for 25% Diuwdowi)

10 219 224 25.7 19.5 25.9 25 9 25.9

2S.9 25.9 2S.9

32.9 32.9 329 32.9 32.9 329 329 329

50 7-6 Ul. 23.6 33.0 38.8 38.8 8.8 38.8 8.8 8.8 38.8 38

75 18.5 20.7 27.4 27.1 2 31.0 32.0 31.0 31.0 32.0 32.0 310

lOO 22.1 ie.3 25.9 2S6 28 5 25.6 2S.6 25.6 25.6 25 25.6 25.6

150 7 1,4 -1.3 7.3 11.4 10.9 10.9 10.9 10.9 10.9 ia9

300 4-3 65 S-6 -3.4

4.4 16-8 14 15.1 15.1



[start] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74] [75] [76] [77] [78] [79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94] [95] [96] [97] [98] [99] [100] [101] [102] [103] [104] [105] [106] [107] [108] [109] [110] [111] [112] [113] [114] [115] [116] [117] [118] [119] [120] [121] [122] [123] [124] [125] [126] [127] [128] [129] [130] [131] [132] [133] [134] [135] [136] [137] [138] [139] [140] [141] [142] [143] [144] [145] [146] [147] [148] [149] [150] [151] [152] [153] [154] [155] [156] [157] [158] [159] [160] [161] [162] [163] [164] [165] [166] [167] [168] [169] [170] [171] [172] [173] [174] [175] [176] [ 177 ] [178] [179] [180] [181] [182] [183] [184] [185] [186] [187] [188] [189] [190] [191] [192] [193] [194] [195] [196] [197] [198] [199] [200] [201] [202] [203] [204] [205]