Mar

26

We've had three weekly reversals in a row recently:

       
           Friday        S&P futures

           2 23          1467
           3 02          1398
           3 09          1417
           3 16          1399
           3 23          1447

… what I would call a positive reversal of a length of three. The weekly moves have been big, approxmately twice as large in absolute value as those of the previous year. What does this portend? It is not above the old highs, and it has gone in a w rather than a v. It has been to Hades once, and then visited it again. Is this a underworldish pattern? Does it continue more after a reversal of length three than a reversal of length one? The two least helpful forecasters I know, one being the columnist who is always bearish, and the other, the ghost who has never yet admitted to an inaccurate prediction, nor made a falsifiable one, are both bearish (of course). How does this change the normal expected distributions? All these ideas have to be tested. 

Alston Mabry comments: 

Pulling out "minimal path" again (min of abs[C_prev-O-H-L-C] or abs[C_prev-O-L-H-C], as % of C_prev) and applying to weeks, to maybe capture more of the intraweek volatility, and then computing z score for each week relative to previous 25 weeks, one does a quick back-of-the-spreadsheet calculation to determine the highest-z weeks since 1990 for the S&P. A recent week comes in at #2:

week of z
27-Oct-97    7.97
26-Feb-07    6.40
28-Mar-94   5.46
19-Aug-91   5.16
15-Jul-96     5.02
30-Jul-90     4.79
22-Jul-02     4.13
21-Nov-94   3.95
10-Sep-01   3.85
31-Aug-98   3.78
3-Oct-05      3.71
3-Jan-00      3.51
15-May-06   3.51
24-Aug-98   3.49
17-Apr-06    3.43
19-Mar-01   3.43
16-Oct-00    3.39
12-Jun-06    3.39
20-Aug-90   3.38
17-Jul-95     3.35
4-Mar-96     3.28
6-Apr-92     3.26
10-Jun-02    3.19
5-Jun-06     3.19
16-Feb-93   3.15
1-Jul-02      3.11
13-Aug-90   3.07
24-Jun-02   3.02

(This list is all weeks with z of 3 or greater)

Where was the S&P six weeks later? Can't know that yet for Feb 07, but here are the z scores (relative to previous 25 weeks) for the S&P's moves six weeks later for the other weeks from the list above:

week of z
27-Oct-97     +0.39
28-Mar-94    +0.21
19-Aug-91     -2.46
15-Jul-96      +0.59
30-Jul-90      +0.84
22-Jul-02      +1.46
21-Nov-94    +0.67
10-Sep-01    +2.02
31-Aug-98    +1.85
3-Oct-05      +1.42
3-Jan-00      -1.51
15-May-06   +0.09
24-Aug-98    -0.43
17-Apr-06    -1.77
19-Mar-01   +2.76
16-Oct-00    -1.28
12-Jun-06    +1.12
20-Aug-90   +3.19
17-Jul-95     -1.24
4-Mar-96     -0.21
6-Apr-92    +0.18
10-Jun-02   -2.03
5-Jun-06     -0.29
16-Feb-93   +0.01
1-Jul-02      -0.17
13-Aug-90  +1.05
24-Jun-02   -0.53

Average: +0.22
Count:     27
Positive:  16, or 59.3%

This doesn't look much different from LT positive drift.

 


Comments

Name

Email

Website

Speak your mind

1 Comment so far

  1. Bryan Franco on March 24, 2007 7:49 pm

    Why was six weeks used as the lag between an absolute z score of three and the price return on the SP500?

Archives

Resources & Links

Search