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]


119

managers and speculators. It is perfectly sensible to look for a pattem in the way many of the long-tenn investors set positions at the beginning of the year, the result of a reallocation of their portfolios, or resetting positions liquidated before the end of the year for tax reasons.

If January is a leading indicator of stock market movement throughout the rest of the year, there is a combination of pattems that should be considered based on the few days immediately after the year begins, and the net maiket direction for the month of January Using the Dow Industrials as a stock market indicator, the January pattem was viewed from 1900 to 1989 in two parts, 1900 to 1937 and 1938 to 1989. The results are shown in Figure 15-8 for all years and Table 15-10 for the past 52 years.

Figure 15-8 shows that the early part of the century had no significant pattem. There were nearly the same number of buUish and bearish indications resulting in a larger number of incorrect predictions. The past 52 years are very different, with a much larger number of correct moves. One should note, however, that the ratio of up to down moves is more than 7: I. During the past 10 years it is Hkely that this pattem would continue to be successfiil if the prognosis was a bull market. To offset the market bias, it should be noted thai the pattems shown in Table 15-10 are very logical. The direction of the first 5 days, confirmed by a continuation in that pattem for the balance of January, is then followed by the same pattem for the year, in the case in which the maiket changes direction after the first 5 days and nets a loss for the month, the pattem conforms to the new direction for the balance of the year.

Reversal Pattems

The last of the three studies looks at reversal pattems. The intention is to find a pattem in the open, high, low, and closing prices of the day that will help predict the next days pattem or direction. Three combinations are tabulated here:

Jay Kaeppel, "The January Barometer: Mj4h and Reality" Technical Analysis ofStocks & Commodities (July

1990).

FIGURE 15-8 Results of Januarypattems. uilts

January up «times Feb-Dec

Source: jay Kaeppel, The January Barometer: Mj4h and Reality," Teclmical Analysis of Stocks & Commodities, 8, no. 7 (July 1990). > 1990 Technical Analysis, Inc. Used witii permission.

TABLE 15- 10 The January Barometer Pattems, I938-I989



S&P 500

During Fira S

By Ihe End

Expectaaons

Pattem

Days of thereof

of January

forFeb~Dec

Declines

Purser decline

Bearish

Declines

Less decline, but loss for tiie mondi

B«ansh

Declines

Gain for the month

Bullish

Advances

Furriier gains

Bullish

Advances

Less gain, but gain for the month

BulHsh

Advances

Loss for die month

Bearish

[ntraday trend continued ("Trend). A continued upward trend is set up when yesterdays high is greater than the pre-ious high and yesterdays close is greater than the previous close. We then find the percentage of dajs that open higher or close higher than yesterday. Continued downtrends are the opposite pattem.

Reversal day ("Reversal). Rtyinning with yesterdays pattem of a higher high but a close below the prior close, a reversal day opens lower or closes lower than the previous close. An upward reversal starts with a lower trend day

Extreme reversal ("Extreme). Yesterdays high was greater than the prior high, but yesterdays close was below the prior low. The reversal fi-om up to down is continued if todays open or close is lower than yesterdays close

Results for these trend and reversal combinations are shown in Table 1.*»-11 for the same six markets tested in the previous studies. In general, the financial markets show few extreme values, while gold and soybeans show a tendency for continued reversals following an upward move. An imusual and very consistent result that appears for all markets is

TABLE 15- 1 IReversal Analjsis for Six Sample Markets,Januarj 1986-Februarj 1996

Market

Pottem

HiihetHuh

Lower Lm

Open

arae

Open

arae

Deutschemarb

Trend

bitreme

S&P SOO

Trend

«

Reversal

bitreme

UST-boiMs

Trend

Reversal

Heuingoi)

Trend

Reversal

Emreme

Gold

Trend

ReversaJ

Extreme

Soybeins

Trend

Reversal

Extreme

the low percentage of reversal follow-through after a lower price move. Readers are encouraged to verify the program printed at the end of this section, to be certain that there are no errors. This imusually persistent pattem may be either a special trading pattern or a program error. If valid, whenever todays low is below the prior low and todays close is greater than yesterdays close, a short sale should prove profitable about SOo of the time.

Progranuning the Reversal Study



eversal Study!

ICopyr.gnt 19 5 1998, PJ Kaufman. A11 rights reserved.

INew reversal study for "Trading Systems end Methods-Each of the following based on direction of current pattern. 1. Continued trend (higher high, higher close, then next open and clo

3. Outside Reversal- Higher high, close below prior low, then ! opi

vars: trend<01, TUopentD), TUclose(O), TDoper[0), TDclOse(D», RUopen{0), RLClose<OJ, RDoperKO), RDclose(O), EUopen{0), ELclO£e(01, EDopen(O) , EDclose(01, nlUlO), nTD<0), nRU(O), nRDlO), nEUlOl, rED(O):

trei.d = close - close[b], Continued trend!

if high[l] > hlQht2] and closetl] > close[2] then begin nTtl - nTU * 1;

if open > closeCl] then TUopen = TUcper + 1; If close > closem then TUclose - TUcl ose + 1, erd;

If lowLi: < low[2: and closeCi: < closet?! then Degln nTD - nTD 1;

if open < closeQl] then IDopen = TDopen + 1, if close < close[i: then TDclose - TDclose * 1; end;

If high[l] > high[2] and close tl] < close[2: then begin nRU - nRU + 1;

if open < closeCl] then RLIopen - RUopen + 1, If close < closeEl] then RUclose - RLclose * 1 end:

if Iow[i: < low[2] and closem > closeCZJ then begin nRD = nRD + 1;

if open > closed] then ROopen - RDopen + 1; if close > Closem then RDclose - RDclose -f 1;

(Extreme Reversal

If hlghm > highCZ: and closeCl] < lowlZ] then begin nEU - nEt + 1;

If Open < close[l] then EUopen - EUopen + i; If close < closed] then EUclose - EUclose * 1; end;

if lov-d] < iDwlz: and closedl > ] then begin nED - nED + 1;

if open > close[i: then EDopen - EOopen + 1; If close > closed] then EDciose - EDclose + 1; end;

If lastcalcdate = date then begin

print ("Total cases:", currentbar;5:0J, print (" open close open close);

pri.it ("Trend ", TUopen100/nTU:4-.D, TUclose.lOO/nTU;4-.0, TDopen-100/nTD;4:0, TDclose-ll

print ("Reversal, RUopen*100/nRL:4:0, RUclose*100/nRL-4-0.

RDopen*100/nTD:4:0, RDclose*100/nTD:4:0, " nRU:4:D. ));

print CExtreire", EUopen*lDO/nEU;4;0, EUclose*100/nEU:4;0,

EDopen*100/nED:4:0. ED(;lose*100/nED:4:0, " {", nEU::0, nEDi-)");

end;

COMPUTER-BASED PATTERN RECOGNITION



[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]