Mar

5

Alexander Elder calls a bar that sticks down (or up) by itself on the chart a kangaroo tail. Price goes to a new level, but quickly pulls back. For example, the second to last bar on the attached chart is a kangaroo tail.

I defined a kangaroo tail as a bar on the 30-minute chart* that had at least 50% of its length either above the high of both adjacent bars or below the low of both adjacent bars. I also required this 50% or greater protrusion to be at least 50% of the average 30-minute range.

I found 18 kangaroo tails in the S&P 500 futures since October 24, 2008, of which 10 stuck down and 8 stuck up. I then calculated the change in the S&P 500 futures from the end of the bar after a kangaroo tail (the earliest point at which a kangaroo tail can be identified) until the same time the next day. Results appeared consistent with randomness:

Close    Close

1 bar     24 hrs

Date           Time  Low          Protrusion Direction     later    later   Change
10/24/2008 9:00   835.0              51% down              860.0 868.0   0.9%
11/17/2008 1:00   850.4              50% down              865.6 831.9 -3.9%
11/20/2008 9:30   774.5              58% down              801.5 747.0 -6.8%
12/31/2008 15:20 890.5              75% down             904.8 920.8 1.8%
1/15/2009 15:00   829.5              60% down              839.3 848.6 1.1%
1/27/2009 7:00     832.6              62% down              840.0 859.3 2.3%
1/30/2009 13:00    823.0              54% down             830.5 815.2 -1.8%
2/11/2009 15:00    826.3              56% down             831.5 835.4 0.5%
2/16/2009 1:00      810.1              70% down             816.5 789.8 -3.3%
2/18/2009 9:30      778.2              50% down             784.0 787.7 0.5%

Average -0.9% Standard deviation 3.0%
N 10
t -0.69
Average of all 24-hour periods -0.2%

Close Close

1 bar 24 hrs

Date            Time      High         Protrusion Direction    later     later    Change
10/24/2008 12:30  882.0       51% up                          860.0    885.0     2.9%
10/27/2008 9:30   881.5         52% up                         871.5    846.5    -2.9%
10/27/2008 10:30  881.0       53% up                           869.0     857.5    -1.3%
11/20/2008 12:00  820.0       72% up                           797.0     753.5    -5.5%
1/9/2009 8:15       915.8        50% up                          890.5     880.5    -1.1%
2/2/2009 15:00     827.5         89% up                          821.3      831.5     1.2%
2/5/2009 11:30     847.5         58% up                          843.5      860.8     2.1%
2/25/2009 11:00   765.0         69% up                           755.5     765.5     1.3%

Average -0.4% Standard deviation 2.8%
N 8
t -0.17
Average of all 24-hour periods -0.2%

* My "30-minute" charts use longer time periods per bar during the overnight session


Comments

Name

Email

Website

Speak your mind

3 Comments so far

  1. Matt Johnson on March 5, 2009 10:28 pm

    Based on your definitions, I would expect a final conclusion/result of randomness.

    You prove that your system, not the tail, is nothing more than random, there’s a BIG difference.

  2. david higgs on March 5, 2009 11:37 pm

    those tail are much better seen in candlesticks…

  3. Jonathan on March 6, 2009 4:17 pm

    If you are looking for non-randomness for a 24 hour return based on a 30 minute bar you are probably looking in the wrong place. It would be similar to trying and predict a month return based on one day…

    Have you considered other holding periods, to see if you can pull any meaning out of your kangaroo tail? Not suggesting that there is, or you can/will, but it might get you closer to something more useful.

Archives

Resources & Links

Search