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]


36

determine the proper stop-loss amount, a computer test was performed using an immediate entrj based on peneU-ation of an extreme boundary with a close stop-loss for risk protection. The first results were thought to have outstanding profits and consistency until it was discovered that the computer had done exactly the opposite of what was intended: It bought when the momentum crossed the upper bound and placed a close stop-loss below the entry It did prove, for a good sampling of commodities, that high-momentum periods continued for enough time to capture small but consistent profits, it also showed that entry rule 1, anticipating an early reversal, would be a losing sU-ategj. One should never forget that declining momentum does not mean that prices are failing.

OSCILLATORS

Because the representation of the momentum index is that of a line fluctuating above and below a zero value, this technique has often been termed an oscillator Even though it does oscillate, the use of that word is confusing. In this presentation, the term oscillator will be resU-icted to a specific form of momentum, or rate-of-change indicator, which is normalized and expressed in terms of values that are limited to the ranges between +1 and -1, +1 and 0, or 100 and 0, as in percent.

To transform a standard momentum calculation into the normalized form (maximum value of +1, minimum value of -1), divide the momentum calculation by the maximum attainable value of the momentum index. A 5-day index for silver with a 20-limit move could be divided by $1.00 to find the normalized value. If silver were to move its limit up for 5 dajs, the oscillator would have the value of+1 rather than the momentum value of 100. If the limits were to expand, the divisor would change as well, giving the technique a means of adjusting for varjing volatility Using the nornralized momentum, or oscillator, the top and bottom zones become volatility-adjusted at any level.

Erratic movements in the simple momentum and oscillator make it a very difficult tool to apply without additional work. Some of these problems can be eliminated by making the bujing and selling zones more extreme or by smoothing the indicator values. In the following sections, a few of the applications of this technique are shown.

Relative Strength Index

One of the most popular indicators for denoting overbought and oversold conditions is the Relative Strength Index (RSI) developed by Wilder. It provides added value to the momentum indicatcr by scaling all values between 0 and 100. It is more stable than momentum because it uses all the values in the period rather than just the first and last, il is a simple measurement that expresses the relative strength of the current price movement as increasing from 0 to 100 It is calculated as

AUthc total of those days dosing higher during the past 14 days az> = the total of those days closing lower during the past 14 days

once the first calculation has been made, both the AU and az> values can be calculated daily using an average-off tnethod:

»j V/etiesmWer. Jr.. New Com:epls in licbnicalladirfs Systems (Ttcndticxtrrh,(jreeBsb 1V78J.

AUtodAUprot - " + todays up close or 0

i4d,„day = dprior + todays down dose or 0

This method essentially drops an average value and adds the new value, if any. The daily calculation of the RSI becomes simple.

Wilder has favored the use of the 14-day measurement because it represents one-half of a natural cycle. He has set the significant levels for the RSI at 30 and 70. The lower level is indicative of an imminent upturn and the upper level of a pending downtum. A plot of the RSI can be interpreted in the same manner as a bar chart, with the



head-and-shoulders formation as the primarj confirmation of a change in direction (Figure 6-6)

In fact. Wilder relies heavily on top and bottom formations for RSI signals. More often than head and shoulders, the failure swing or divergence denotes an unsuccessful test of a recent high or low RSI value.

Modiing the RSI

An obvious objection to the RSI might be the selection of a 14-day half-cycle. Ma.ximum divergence is achieved by the use of a moving average that is exactly one-half the length of the dominant cycle, but 14 days may not be that half cycle. In addition, common-use chart interpretation might mean that the RSI value remains outside the 70-30 zones for extended periods rather than signaling an immediate tum.

In practical terms, a 14-day criterion means that a sustained move in one direction that exceeds 14 dajs will retain a very high RSI value and may result in losses if a short position was entered, if the trading of the RSI seems to have high risk, increase the overbought and oversold levels by moving them from 70-30 to 80-20. This will cause a signal to be given at a more extreme level.

The generalized program code for an n-day RSI is given as

sumup = 0; sundn = 0;

for i = 1 to n begin

if close - closeCi] > 0 then

sumup = sumup + close - closeCl]

else stimdn = sumdn + closed] - close; end;

RSI = 100 - (100 / (1 + (suinup/sumdn)));

FIGURE 6-6 RSI top formation -e70

.Second high farts Sell at break pomt



Ki:uc]> tiy A.I r 1 tin ttit distril-iuticin of clif RSI Klinwtfrd tliat 11 - avc-rac valup of an Rf>l rtip acicl buttum waji c;unMiMt*:nll> (uup*:!] near and " «? petzUve-iy 1h refore of

all KSI values lie bciwercai 72 and 32 wlii h evenl di ti-ibuied and inilar n r»-nrHimon tn a standard deviati™*. This wiuld .l.jys." t dial 70- l<-vtrls i>i-t»j><.>s«.-d by Wilder are

too dose tOKetl-ier to acrt as selective overboitglityoversold values, trut sliould tx.- movcrd farther apiart

I "In- <i • - .-1 at which the mu-ket will i • -i»c.:! i these extreme levels ean be aaju*>te<J by changing the 1-i-day orlteria. If the interval is shorttnetJ, the RSI will reaczh extretnes more often and trades will have higher risk if the tmme pieriod is mcreastzJ, there will be saler but fewer trades based on RSI extremes (discussed in more detail in Chapiter 22>. It in always safer to err on the side of less risk. If there are too many trades being generated by the RSI, a eombinatioii a longer inteial and higher eonfidenee bands wili be an imf-irovemena "*

r SmttoThlng wfth - > L/ps <tncf Downs

InsteucI *jf in* reading the n imbei- of days in the RSI caleulation period, a smoother Indiea- can l>c round i>y m reasiiig Hie period over which each *jf the and dttttn values are

determined The original RSI method uses 1-4 individual days, where an up day is a day in which the price change was positive Instead, we can replace each l-day change with a 2-clay change. Or an »j-day change If wc u e -day ch nges d en a total of ddys will be needed, so that each 2-clay period does not overfap anoth there will be 14 sets of 2 days-LJsing 1 -4 sets of 2 will give a smoother indicator than using 28 single day changes.

Another variation *jn the RSI is the use of the difiference between the sum of the up days and the sum of tl«e down days, called a rtet mf/mt-rttum iscilUtttjir, If y«ju consider the unsmoothed RSI = 1OO x <;.S-„/<.S"„ SYi, then the net momentum oscillator wouid be CMO - lOO X fS - S,,y/(,S -H .Srf> This method replaces some oF the Indicator movement lost to smoothing in the normal RSI and shows more extremes. This may also be done by shortening (he number of periods in the RSI c-,alculacion.

St:ochs(st:ic:s

The stochastic, created by George I-.ane, is an oscillator that measures die relative posiiJon of (he closing price within a past high-low range. It is based on the commonly accepted observation thai closing prices tend to resist penecraring the day s high prices as an upward move ains -trength -Imllarly, clo.slng prices often remain above the lows durinj a declin When tho market is about to trurn from up to down. lor instance It is often the case that the high are h-gher, but the closing price settles within the previous range Thus makes the stochastic oscillator dlfTerent fi-om most oscillators, which are normalized rep-rtsentations of the relatiw-n; str-rgtb, the difference between the closing price and a selected trend speed

The- rhi-ee in«lieatoi-s that result from the stochastic measurement are called 9& and "XJf-slot, The first (wo calculations form the normal concept uf the stochastic oscillator and are c-,alculated for today t as

Initial %«; = 100 X

* Peter w Aan, *How RSI Behaves. Futures Clanuaiy 1985).

* For a more sophisticated approach to RSI optimizatjon, see John F. Ehlers, "Optimizing RSI with Cycles," Tecb-nical Analysis of Stocks & Commodities (Februafy 1986)

Tushar S. Chande and Stanley KroU, Nae Tedmiail Tntter (John Wiley & Sons, New Ybrti, 1994).

where C, IS todays closing price

1,(5) is the l(jw price of the last 5 days

K,15) IS the range of the last 5 days (highest high minus lowest low) as of "

Notice that "cD is a 3-day moving average of "cK, and "o/.D-slow is a 3-day moving average of "oD. The 5-day high, low, and range selections tend to make the smoothed stochastics, "oD and "oD-slow, front-loaded calculations. Figure 6-7 shows the price of July 198-1 cotton plotted with the corresponding "oD-stow and "oD (°oK-slow) stochastic. Notice that the extremes do not correspond to what would be expected using a 5-day simple



[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]