One is always fascinated by a beautiful femme fatale that always keeps you guessing and never does the same thing twice, sort of like Federer's new drop shot or the market mistress.

To test whether there was such a thing as dueness, I had to start somewhere. Do the four urns get filled with balls of different colors to an inordinate degree? Here's a start:

Table I

Number of occasions that a small change failed to occur, by days of failure and probability of continued failure next day

Small rises

#/Observations/Prob. a failure occurred the next day

0 507 0.77
1 388 0.77
2 296 0.82
3 243 0.77
4 187 0.80
5 150 0.83
6 124 0.81
7 100 0.81
8 81 0.91
9 74 0.88
10 65

Table II

Small Declines

0 396 0.83
1 330 0.83
2 275 0.85
3 234 0.83
4 194 0.83
5 161 0.82
6 132 0.82
7 108 0.85
8 92 0.84
9 77 0.80
10 62

there were 2600 days in sample. A small change was defined as less than 0.5% and there have been four such changes in a row.

For example there were 507 days that a small rise occurred; on the next day there were 388 days a small rise did not occur (i.e. 119 where a small rise did occur). Note that 388/507 = 77%.

In conclusion, one will have to reflect on other methods of teasing out this mysterious woman's ability to deceive, possibly with your ideas. 

Steve Ellison submits his study:

After a certain number of whipsaws, is a strong trend due? I tested S&P 500 data from 2005 and 2006. I defined the trend as the direction of the last move greater than or equal to the average true (daily) range. By this definition there is always a trend.

I started with a hypothetical always-in mechanical strategy that one would be long whenever the trend was up and be short whenever the trend was down (I realize real traders don't trade this way). I then categorized the results of each hypothetical trade into the 20% that were most profitable and the other 80%.

I then counted how many of the previous 10 trades would have been in the top 20% and tabulated results by this quantity.

Number of top 20%       Average

trades in previous 10    profit     N

.       0                  -0.1%    20

.       1                   0.0%    44

.       2                   0.0%    57

.       3                   0.0%    55

.       4                  -0.3%    19

.       5                  -0.6%     3






Speak your mind

5 Comments so far

  1. douglas roberts dimick on June 9, 2009 10:38 pm

    For purposes of linear regression, may one conclude that the model is balanced based on the common range of sums (rounded down 2100/2200 respectively) of the matrix rows?

    Albeit factoring probability of failure, the theory that markets channel approximately 80% also appears here given that ranging (2100/2200) within the “bivariate generating function” less the days (subtracted from 2600) so implied without small rises and declines — inferring direction force.

    Query: Why does the failure probability progressively increase with small rises and not declines?

    We assuming here the 2600 days are consecutive and from present to past, 252 trading days a year… so the last 13 years?

    Alas, my rules-based limitations here evince poor math… dr

  2. Craig Bowles on June 10, 2009 8:58 am

    We're in the same odd place as early 2002 where leading economic indicators are improving but inflation indicators are running up almost as fast and not showing the three-month lag anymore. So far inflation indicators aren't outpacing yet but the 2002 price action for stocks moved up to new highs in June and then sees the S&P trading back off to 875 in July. The four-month cycle suggests the next low is late November to early December, so some sort of late July low would be in line. All the government actions are confusing but the basic economic cycle rules are working as normal. The only real surprise is gold normally would be getting killed, so holding up so well makes you wonder what happens mid-cycle of the next economic expansion. If you're frustrated by recent govt policy, check out this guy!

  3. Laslo Minks on June 10, 2009 5:22 pm

    In the same spirit: what tends to happen when highly correlated markets decouple? Something non-random?

    I see markets like romantic relationships. Breakups are often ugly. Oil has pursued the S&P the past year with flowers in hand, but does the S&P really like Oil? Will she put up with Oil's constant advances? If S&P rejects Oil, how hard will Oil take it? Has a weak USD played matchmaker? If things don't go well, will the matchmaker have to choose one or the other to remain friends with? Will other friends and acquaintances have to choose sides? Will those with mutual exes form new camaraderies? Are there celebrity relationships in the markets, which other markets live through vicariously? If the prince and princess break up, will the whole world be forced to choose one side or another?

  4. Laslo Minks on June 10, 2009 5:36 pm

    Of the above, the hypothesis I am most interested in: Will those with mutual exes form new camaraderies? If so, is there a consilience with Is It Due?

  5. vniederhoffer on June 11, 2009 7:19 am

    May I compliment Mr. Minks with some spotted dog or figgie duffy for one of the most creative posts here in 10 years and 15,000 voyages. If one weren't a yellow admiral, one would haul out the silver. Vic Aubrey


Resources & Links