close, indicates that it has much more noise than Deutschemaiks and most other markets. A unique pattern of the S&P is that large upward gaps alwajs trade lower while large downward gaps have occasions when prices just keep failing.

Treasurj Bonds

This market, as with the S&P, has had an upward bias during the past 10 years; therefore, the sjmmetrj of the pattems may be affected. For bonds it is interesting to view the small gap open, for example .125 equal to 4/32. There is an 86**o chance that prices win cross the prior close during trading, but a much smaller chance (38**o) that prices win close lower. The 52**o chance that prices wiH close above the open creates an opportunity for bujing, provided risk can be controlled.

Heating Oil

Expressed in cents per gallon (.020 is 2 cents), this study shows a large number of extreme gaps exceeding 2 cents. When those gaps are higher there is a strong tendencj to continue on the next open.

Soybeans

The extremes of soybeans, as with some other markets, show a tendencj to continue the next day Large gaps down are more likely to be consistently lower, while upward gaps are less consistent during the day.

Gap Study Program

The following Omega Easy Language program is easuy converted to other programming languages. 11 can be used to e\aluate gap opening of any market based on dauy prices. The results are printed to a flat file called "gapout" stored in the Windows directory. Note that the results are given as the number of cases satising each condition rather than the percentages seen in Table 15-6. To print the percentages, simply multiple the value by 100 and divide by the number of total cases (nit ems) in the print statement.

iSVbTEM: Cap Study)

(Copyright 1978 1998. PJ Kaufnen. All rights reseved.l Inputs. 1 ( (0), (20);

vars. r(0!, r2(D), RiirO), BpD,. ;0), ii . ),

arrays- Traxmin:bO]CO), tcont [bO]{0). cbf 1 cwO] (!). <rev[0](ll) -rcrcssCeCJ XrstijihDMO, trrnntLfiCld;;, rr.nrt[M].:();, r-tprisLf.r 0., lliitid-.ize ircrements)

i <= 1 Iher begir r = rbars + 1; r\Z = @intport ior(n/?}, middle = 2 + 1; mil - -n2*gflpircr; for ix = t to begin

mxniln[ix] = m1r; mir = mm + gapincr; crd;

erd;

IProcess gap!

if open <> closed] tnen begin

gap = (open - IoseLH)/gapircr, if gap > 0 tnen begin

ngep = gintportiontgep) + middle + 1.

if ngap > n tlicn ngap - n:

end;

1f gap < 0 then begin

ngap = middle - 1 + @intportior,{gapi; Igap is ncgative,

if ngep < 1 then ngap = 1;

end;

ntemsCngap] = nitem5[nqe(i] + 1; IRelative position of todays close!

if {gap > and close > open) or (gap < 0 and close < open) the"

ccontfrgap] = front[ngap] + 1; if igap > 0 and close > closed] and dose < openj or

(gap < 0 and close < closeLU and close > open) tnen

cbelDw[ngap] cbelow[ngep] + 1; if (gap > 0 and close < closed]) or (gap < 0 and lIlsp > l;fsp"l]> hen

crev[ngap] = crev[ngap] + 1; (Pattern of entire trading range)

if (gap > D and low < closeLU) or (gap < 0 and high > closed]: then

trcrossLngapJ = trcrossLngap] + 1; if {gap > 0 and low > clOse[l] and iow < open) or

(gap < 0 and high < close[l] ard high > oper) t4en

tradjLrgap] = lradj[ngap] + 1; if (gat) > 0 and low open) or {gap < D and high <= open) then

trcont[ngap] = trcont[ngap] + 1;

ICortinue next day)

if(gapd] > 0 and open > close[l]) or

(gapd] < 0 and open < closed]) tlier ncont[ngap] = ncont[ngap] i- 1;

else

ritems[middle] = rn tpmsLniddle] + 1; if lastcalcdflte = date then begin

for ix = r downto 1 begir

prirt ( file(C: Windows\gapDut"), niaxminl ix] :4.3. riitemsLixj:5:0, ccont[ix]:b;0, cbelow[ix];4:0, crev[ix];4:0, trcr0S£[ix]:S.0, tradj[ix]:4:D, trcDnt[ix]:4:0, ncontLix]-5:0); erd;

end;

THREE STUDIES IN MARKET MOVEMENT-WEEKDAY, WEEKEND, AND REVERSAL PATTERNS

The next three sections are concerned with longer periods of time. The first, weekday pattems, looks only at the closing prices during a 5-day week fi-om Monday through Friday to find recurring closeonly pattems. A week is ofl;en considered to have integrity as a single imit of time, and participants are accused of acting in the same way each week. This study attempts to isolate predictive patterns., for example, if Monday through Thursday were all higher, what is the possibility of Friday being higher? if the weekly trend can be clearly identified, should a correction on Friday be expected due to evening-up (closing out positions before the weekend)? Because pattems represent himian behavior, the results could prove interesting.

Weekend pattems are considered independently in the second study. The opening direction on Monday could restore the tiend of the prior week, if those traders who liquidate on Friday intend to reset their positions. The weekend is also an extended period for imexpected news or a buildup of public interest. The study win attempt to relate the direction of the Monday opening price to some pattem or tiend of the preceding week.

Reversal pattems are not based on a day of the week but have often been discussed as a leading indicator. They may also be used as a filtering or timing dee.

Weekday Pattems

During their constant exposure to the market, professional traders often observe pattems in weekly price movement; their acceptance of these pattems is as old as the market itself In Reminiscences of a Stock Operator, the fictional character Lairj Li\ingston (assumed to be jesse Livermore) begins his career recording prices on a chalkboard above the floor of the New York Stock Exchange, eventually becoming aware of pattems within these prices. The most accepted occurrence of a pattem is the Tuesday reversal, which is taken as commonplace by close observers of the maiket. When questioned why a stiong soybean maiket on the first of the week is followed by a weak day, a member of the Board of Trade would shrug his shoulders and quote: "Up on Monday, down on Thesday" if this is true, there is a trading opportunity

if a commonly accepted idea is not enough to be convincing, consider the additional rationalization about human behavior: The weekend allows a buildup of sentiment, which should result in greater activity on Monday. Coupled with adding back positions that were liquidated prior to the weekend, this may cause a disproportionate move on Monday, especially early in the session. This pattem may be fiirther exaggerated when a clear tiend exists. With this overbought or oversold condition, it is likely that Tuesday would show an adjustment. So much for hypothesizing

The first aspect of the test was to define the weekday pattem. This was done in terms of the Friday-to-Monday move (close-to-close). Monday alwajs received the value X, regardless of whether its direction ftom Friday was up or down. For each day that closed in the same direction as the Fridajto-Monday move, another X is used; when the close reversed direction, an 0 is recorded. Therefore, XOXXO means that Tuesday and Friday, represented by 0, closed in the opposite direction ftom the prior Friday-to-Monday move, while Wednesday and Thursday were in the same direction. This could have meant either of the situations:

(I) (2)