Wednesday, August 26, 2009

Weekly Outside Bar and NR6?


What about NR6s?


My internet service was unexpectedly down for the last two days, hence no updates...Never in my life have i had to work so hard to give someone my money! That really blew! Anyway, I have one short term and one interim term study to share. Both show a negative bias, but also indicate limited down side.


You have probably heard of an NR7 day. For those that are not familiar what that means, it is nothing more than short hand for a day's whose range (high - low) was the narrowest of the last 7 days. Well we did not get such a day today, but we may tomorrow...that remains to be seen. What about an NR6 day - is there any edge that it can offer? Well it turns out there is.
Today's range was the narrowest of the last 6 while the S&P made a new high. In the past decade, there have been 40 such instances. In 34 cases (or 85% of the time), the market was trading lower within 3 days.

TickerDate/TimeClose3Day-%Chg N1%Chg N2%Chg N3
^GSPC3/8/19991,282.731.00-0.230.321.17
^GSPC4/26/19991,360.042.000.20-0.67-1.27
^GSPC5/13/19991,367.561.00-2.18-2.05-2.50
^GSPC7/2/19991,391.221.00-0.220.330.23
^GSPC7/7/19991,395.861.00-0.100.530.23
^GSPC7/9/19991,403.281.00-0.30-0.69-0.36
^GSPC12/22/19991,436.130.001.551.461.50
^GSPC12/29/19991,463.463.000.070.40-0.56
^GSPC3/22/20001,500.640.001.781.791.55
^GSPC6/3/2003971.560.001.511.911.67
^GSPC6/17/20031,011.661.00-0.16-1.68-1.58
^GSPC10/31/20031,050.710.000.790.240.10
^GSPC12/12/20031,074.141.00-0.570.090.22
^GSPC12/17/20031,076.480.001.181.131.53
^GSPC12/23/20031,096.021.00-0.18-0.011.23
^GSPC12/30/20031,109.642.000.21-0.101.13
^GSPC12/23/20041,210.131.00-0.430.280.27
^GSPC3/7/20051,225.311.00-0.48-1.49-1.31
^GSPC8/3/20051,245.041.00-0.74-1.50-1.76
^GSPC11/25/20051,268.251.00-0.85-0.85-1.48
^GSPC1/9/20061,290.151.00-0.040.31-0.32
^GSPC3/17/20061,307.251.00-0.17-0.77-0.17
^GSPC4/5/20061,311.561.00-0.19-1.22-1.14
^GSPC5/2/20061,313.211.00-0.39-0.070.96
^GSPC9/27/20061,336.592.000.17-0.06-0.39
^GSPC9/28/20061,338.881.00-0.23-0.56-0.36
^GSPC10/10/20061,353.421.00-0.260.700.90
^GSPC10/24/20061,377.383.000.350.85-0.00
^GSPC11/21/20061,402.812.000.23-0.13-1.49
^GSPC12/5/20061,414.761.00-0.13-0.53-0.35
^GSPC1/12/20071,430.732.000.08-0.01-0.30
^GSPC1/16/20071,431.901.00-0.09-0.39-0.10
^GSPC2/2/20071,448.391.00-0.10-0.030.11
^GSPC2/15/20071,456.811.00-0.090.200.06
^GSPC5/7/20071,509.481.00-0.120.21-1.19
^GSPC5/21/20071,525.101.00-0.06-0.18-1.15
^GSPC5/31/20071,530.620.000.370.560.02
^GSPC6/2/2009944.741.00-1.37-0.24-0.49
^GSPC6/12/2009946.211.00-2.38-3.62-3.75
^GSPC7/27/2009982.181.00-0.26-0.720.47
^GSPC8/26/20091,028.120.000.000.000.00

Note that while the bias is clearly negative, except for the last few instances, generally downside has been limited.


Weekly outside day at new 10 week high

Last week the S&P the S&P painted an outside bar on the weekly chart while closing at a new 10 week high. There have only been 12 similar instances in the past decade. As you can see from the table below, the market closed lower in all cases but one within the next 3 weeks. However, observe that in the past 3 years, the declines were less than 1% were you to exit on the first weekly lower close.

TickerDate/TimeClose3Day-%Chg N1%Chg N2%Chg N3
^GSPC1/8/19991,275.091.00-2.50-3.910.36
^GSPC4/23/19991,356.851.00-1.60-0.87-1.40
^GSPC12/3/19991,433.301.00-1.13-0.861.75
^GSPC1/4/20021,172.511.00-2.30-3.83-3.35
^GSPC12/12/20031,074.140.001.352.023.20
^GSPC10/1/20041,131.501.00-0.83-2.06-3.16
^GSPC1/6/20061,285.452.000.17-1.86-0.13
^GSPC4/21/20061,311.281.00-0.051.10-1.53
^GSPC9/15/20061,319.661.00-0.371.232.27
^GSPC1/12/20071,430.731.00-0.02-0.601.23
^GSPC2/16/20071,455.541.00-0.30-4.70-3.62
^GSPC4/18/20081,390.333.000.541.70-0.15
^GSPC8/21/20091,026.130.000.190.190.19

3 comments:

Blue said...

These seem like strange studies. Consider the first study. If you looked at all days (instead of just NR6 days), in how many of them would the market have finished down at least once during the following three days?

Are NR6 days really that different from other days? If so, it would have been a good idea to show how different they are. Even better would be a formal measure of the statistical significance of that difference. Otherwise how can you say what this data really means?

The Small Fish said...

hi Bigblue, good point. Generally i do not post or consider studies that show odds 70-75 pct odds. That is because for most of these studies, my time frame is 3 to 5 days and for that time frame, i know believe the odds are in the low to mid 60 percentile (i can not check at the moment as i am in transit). I used to post the 'anytime' odds for comparison, but given that the bulk of these studies use similar time frames, i haven't recently. Thanks for commenting on this - ill try to include that in the commentaries going forth.

The Small Fish said...

Blue, sorry bout' all the typos and calling u 'bigblue'..guess my head is on techs and im realizing my fingers are too fat for this windows mobile device.

Post a Comment