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]


168

used. There may be an advantage to the double moving average approach, but a third, fourth, and so on, lose their significance. A momentum indicator bears a close resemblance to a moving average when its rules are to buy or sell in the direction of the price move, especially when it crosses the zero line. Both have great similarities and will not offer much diversification when applied to the same data. Each sjstem building block should be fundamentally different. A technique, such as a breakout method, that does not depend on time would be a sensible complement to a moving average, which is driven by time. The more complicated cycle and wave theories, contrarj opinion, and other methods all view price movement in a completely different way than a moving average, even when applied to the same data The same moving averages may make the most sense when you apply the same period to different tjpes of data, such as volume and open interest, or compare two related series, such as the Swiss franc and German bund exchange rates.

Filtering is a tool not reshicted to price analjsis, but applicable to money management, reliability, ride portfolio selection, and other aspects of a trading program. Selection of which markets to trade and position size is as important as the buy and sell signals. If the sjstem itself corrects for high-volatility markets by decreasing the distance of the stop-loss point (keeping the risk constant), it would be redundant to trade fewer positions of that market, thereby reducing the effects of any profits or losses relative to other iterns in a portfolio.

SYSTEM TRADE-OFFS

Although traders are continually searching for a sjstem that has no losses, they are realis. tic enough to know that most profit, loss, and rid; decisions are dependent on one another, just as in Gambling Theory a sjstem can be structured to have frequent small losses and occasional large profits-or frequent small profits with some large losses. It is simply not possible to create a sjstem with such small losses that these losses are essentially zero without increasing the frequency of the losses to the point at which profits rarely occur. In short-term or high-leveraged frading, the cost of doing business prevents such an approach from working.

The basic principles of sjstematic frading require that the trader choose between the frequency of profits to losses, the relative size of those profits and losses, and the number of opportunities that will be available to trade. These interrelationships will be different for various tjpes of sjstems.

Trend-Following Sjstems

The relative rids and rewards of a frend-following sjdem are based primarily on the selection of the trend speed. As an exanple, using a simple moving average sjstem applied to closing prices. Figure 22-7 shows that a faster moving average (smaller number of dajs) will stay closer to the current price than a moving average that uses a larger number of days. The maximum risk is equal to the largest lag, as measured from the current price to the corresponding moving average value. This lag is limited by:

M=p-axK)/2

where P is the current closing price, L is the limit move or 3 standard deviations of the average daily frue range, and N is the number of dajs in the moving average.

The smaller losses of the fader frend sjstem are kept in proportion to the profits because of the relative sjstem sensitivity. The price changes of smaller magnitude, which cause the moving average to change direction and hold losses to a minimum, also break the sudained trends into shorter movements. The net results are more frequent trades with relatively small losses and modest profits. Longer-term trends cause fewer trades by

FIGURE 22-7 Relative risk of a moving average sjdem.



holding to the maifcet direction tor a longer time. This results in larger profits and correspondingly laiger losses.

The risfc of a single trade being small or large is not a complete measure of trading risfc. Because there are more losses than profits in normal trend-following sjstems, it is important to consider the sequence of losses as part of the ultimate risfc of trading. It is the aggregation of these losses that provides a reasonable comparison in ddermining whether a slow or fast moving average sjstem has the best profit to loss ratio.

The feature of a trend sjstem that is most sppealing is the small losses relative to large profits. This feature also causes frequent small losses during sidewajs marfcets, and the number of these false signals caimot be significantly reduced simply by using a longer-term frend. Testing will show that there is always some sidewajs price movement that coincides with the frendline, resulting in a high percentage of whipsaw losses. It is unrealiatic to expect a smoothing technique to produce more than 40°o to 45" profitable trades.

Countertrend Sjstems

The rid; and reward sfructure of a countertrend method is different from that of a frendfollowing spproach. Although the size of your profits and losses relates to the time period of analjsis in a manner similar to frend following, positions must be set in advance of an expected price reversal. This clearly infroduces more risfc and makes the success of the method dependent upon good timing. When countertrend methods are discussed here, those techniques thai require a confirmation (i.e., they wait until prices reverse their direction) are not included as countertrend sjstems

Due to the greater rid; of a countertrend sjstem its objectives are generally well defined. In a frend sjstem you let the profits run until prices reverse; however, most countertrend methods rely on an oscillator, contrary opinion, or an overbought/oversold analysis, and set goals for exiting when the overbought or oversold condition dissppears. This results in a profile that limits profits; therefore, it should not also limit losses.

Trade-Qffs

The selection of a performance profile goes together with particular frading stjles. Some combinations are entirely incompatible. Ideally, you would want very large profits, high reliability, and small losses. Unfortunately, these combinations are not possible. The tradeoffs that are readily available are:

1. A few large profits and frequent small losses

2. Many small profits and afew large losses

3. More opportunities with greater risk

4. Fewer opportunities with less risfc

It is easj to identify the first combination with frend following. To have fewer losses you must allow greater rid; Keeping losses small means being stopped out because of marfcet noise rather than underljing frend; therefore, you can expect many more losses than profits. In the case of a countertrend sjdem you might profit from maifcet noise, which accounts for a large percentage of the price variability. The moves caused by maifcet noise are also much smaller and



more frequent than trend moves. To profit from noise you must enter the martlet while prices are going in the opposite direction, as in a countertrend technique, and allow more rid;. This causes the profile in option 2, many small profits and a few large losses.

The last two choices are the result of a random dishibution of price moves, if you look bad; at the study of runs, you will remember that price changes exhibit a pattem very similar to random movement. For every 100 dsjs of prices, there will be 50 dsjs in which prices

reversed after only 1 dsy of up or down movement, 25 dsjs of price movement in the same direction for 2 consecutive dsjs, 12.5 dsjs for 3 consecutive dsjs, and so on. Although real price pattems may varj slightly from that dishibution, they are very close. For any fixed time period, these pattems remain the same; that is, there are alwajs more opportunities for smaller moves than larger ones. To hold a frade for the long term, you must be able to withstand all of the market noise that might occur during that period. The longer you hold a trade, the greater chance you will see larger market noise, or adverse price moves.

You can see how this spplies to a countertrend trading using a tjpical price oscillator. Figure 22-8 shows the fluctuations of a sidewajs market, one that has no frend bias. Assimiing that there is a normal disfribution of prices there is a clustering near the center and less frequent peaks and vallejs as prices move fiirther away from the middle Consfruct an overbought/oversold indicates, which can be categorized as slightly (S), moderately ( ), or exfremely (E) overbought or oversold, by drawing lines on Figure 22-8.

It can be seen that there are more opportunities to buy a slightly oversold maiket and sell a slightly overbought maiket; all prices must pass through the area closest to the center before reaching a more overbought or oversold condition. Therefore, the opportunities must increase as profit goals decrease. But what about the rid;? If profits are limited, risk must be reshicted to a small part of the potential profits, or not limited bv a dollar amount but by a pattemed condition. In this counterfrend sjdem example, an attempt to make losses small relative to profits entered during the slightly oversold zone would cause the number of losses to increase dramatically, overwhelming the potential profits (seen in Figure 22-8). If losses are not reshicted, they will be laiger when there are more opportunities with smaller price objectives, and smaller for the fewer cases when prices are in the exfreme zone. The minimum risk occurs on a single trade taken at the absolute extreme points. Figure 22-9 shows the relationships that exist in a standard countertrend sjstem.

All sjdems have frade-offs. To reduce the losses, either the profits or the frequency of opportunities must be reduced whether the frading spproach favors the frend or counterfrend. A frend follower looking for the big move musl also take a big rid;. In the development of a sjstem or frading philosophy, personal preference determines the combination of rid;, reward, and opportunity.

The Commodex Method of Combining Elements

The Commodex sjstem was first presented in 1959 and is atill active, published daily by Commodity Futures Forecad in New York. Commodex combines the components most 1 1 to the experienced trader in a unique weighting method. It includes moving averages, price momentum, and open interest to calculate a frend index.

The most interesting aspect of the Commodex sjdem is its ranking process, intended to produce a relative sfrength value for each frend. Using a 10- and 20-dsy double mo .

FIGURE 22-8 Counterfrend alternatives using defrended price series.

FIGURE 22-9 Relationship of profits to risk per frade based on opportunities.



[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]