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]


24

Smoothing ReskJuol Impact

2/(r+ 1)

10%

.333

13.17

.369

.451

.182

13.44

.259

.125

13.49

.181

.095

13.51

.109

.139

where M is the level of residual impact (e g., .05,. 10, .20). A lower percent level impUes more residual impact, and n is the equivalent number ot moving average days.*

The approximation for the smoothing constant, given in the previous section as 2/{n+ 1), can be shown to have a tmsisiMit residual impact ot between 13 and 14%. The use ot the preceding formula, as shown in Table 4-9. would allow the specific adjustment of residual impact.

DonaldE Lambert. "Ezfonentially £n,o..fl,edM..n„g Aver-es." TecWcal Analysis ..f . .. & Citfi.odities (ceftemberiIttober 19iM)



5 Trend Systems

The previous chapter developed the tools for calculatmg trends-a standard movmg average, weighted averages, exponential smoothing, and regression-based values. To profit fran identifying the trend requires the use of trading rules and the selection of specific parameters that define the trend speed and position rislc among other factors. Often, the selection of rules is seen to be dependent on additional parameters. This chapter will first discuss those rules that are necessarj to all trading strategies, then give examples of actual sjstems. The selection of trend speed is handled only briefly here, but it is continued with a detailed analysis of these and other sjstems throughout the booic. and especially in Chapter 21 ("Testing"). Toward the end of this chapter, a section on viewing sequences of trends may prove interesting to the analjst trjing to be more innovative.

BASIC BUY AND SELL SIGNALS

of the selection of moving averages, the smoothed trend calculation >o-,-in lag behind the actual current market price. In rising markets, this lag will cause the moving average to be below the price; in falling maikets, it will be above. When prices change direction, the trendline and the actual market prices cross because the trend is slow to reflect the new market direction. These crossings provide the opportunity for the basic trading signals

Buy when rising prices cross the trendline.

Sell when declining prices cross the trendline.

It is also necessarj to decide what prices are to be used. The answer depends on the construction of the average and the testing of the sjstem. If the average is composed only of closing prices, it is reasonable to start testing with the modified signals

Buy when rising prices close above the trendline. Sell when declining prices close below the trendline

If an average of high and low prices were used, or the high. low. and close, a buy or sell would be signaled when the new average (high + Iow)/2 or (high + low + close) 3 was above or below the corresponding moving average value. In all cases, consistency is important. The sjstem that is tested and the one that is traded should be the same. In the following discussion, the term close will be used. To find out if a trading signal occurred on any day, wait until the close of trading, calculate the new moving average or trend value, and then see whether a crossing occurred according to the basic buy and sell rules.

You may have realized that using the closing price to generate a signal usually requires that calculations occur after the maiket has closed, and orders must be placed in the afteiiiours maiket or on the next days open, if the sjstem is actually generating signals as of the closing price, then you are not following the sjstem. To correct this problem, the value of the closing price that would cause a new signal should be calculated in advance. This would allow execution on or near the close of the current day. The rules of the sjstem can also be changed to buy or sell on todays close if that price is above or below yesterdays

moving average value. A small amount of testing will determine which approach woiks best. Another choice would be to test todays close against yesterdays projected moving average; therefore, if the moving average values for the past 2 dajs were Mt-i =432 and Mt.2=29, the difference Mt-i - Mt.2 would be added to Mt-i to get the estimated fuming criterion of Mt for today. That means that the rede of cfiaiige must differ for a signal to oocur. More sophisticated approaches to projecting the moving average will be covered later.



Variations inTiming

Lead and lag are features common to the trend calculations discussed in Chapter The opposite in purpose, a lead of n days advances the plotted value oi, for exampie. the moving average M, firom the same lime as the current price P, to the position n periods ahead In actual trading, this gives an advant:c- criterion for a trading signal; prices must remain above this level to maintain the current position. A lag time of n days plots the current moving average value at the same time as P. „, n days sooner. This indicates that a signal has already occurred, but no action may be taken for days The lead plot tries to anticipate a trend change, whiie a lag is a commitment to holding a position. Both lead and lag techniques may use the same basic buy and sell signals as standard trends.

Entry and exit timing may be improved bv a delay in taking action on a signal, usually from 1 to 3 days. This technique allows a number of days for the newly developed trend to reverse and show a false signal at the cost of missing the beginning erf" the trend. S\-stems that have a large number of short-lived losing trades and a high profit-loss ratio can benefit from a delay of 1, 2, or 3 days

Reversing a position on each trade signal results in a constant commitment to being in the market This philosophy will show greater profits when the trade is successful, but more fi-equent losses because of the constant commitment to trend changes during whip-saw periods. An entry delay may also improve a reversal system

BANDS AND CHANNELS

Thus far, modifications have been based on a simple moving average or trendline- The plot can be acjusted forward or backward, the position can be reversed, or the entry or exit delayed in an attempt to begin the new trade at the time when the trend has decidedly taken a new direction. Even if prices have begun what will become a majoi downtrend, an entry into a short position too soon may be subjected to sharp reversals caused by conflicting fundamental and technical elements at these turning points. The simplestway to avoid these price variations is by using a band.

A band is an area surrounding a trendline above and below that acts as a zone of commitment for the trader and allows time, as measured by price movement and risk, for prices to setUe into their new direction. Also called channels, bands can be created in many waj-s, one of the most popular being a percentage of the current price or the current trendline value. A band that is formed fran the trendline will be

(upper band) B,; = + cM, (lower band) - ,- cM,

where is a percentage, 0 < < 1, and M, is todays moving average value. Since the moving average is much smoother than the price patterns, the band will be fairly uniform as well. If the current price is used instead of the mo\dng average then

B=M, + cP,

The band will wary in width at a much faster rate using price. During an uptrend, the band based on prices can get very wide. Another popular type of band is based on absolute point value, for example, silver 50, com 50, or catUe 25 points; eadi represents a dollar rid; to the trader. The use of this tjpe of band is usually found coordinated with a money management program that limits losses on trades to a specific number or to a maximum percentage of the portfolio being managed.

The independent smoothing of the high and low prices for any period forms a volatility band. Although it may be practical to use the same technique or the same relative smoothing (e.g, 10-day or 10°o smoothing constant) for the hi, low, and closing prices, it is not a requirement, if the same smoothing criterion is used, the band will be uniform



[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]