|
PureBytes Links
Trading Reference Links
|
Here are the results of using a 10% trailing stop filter on the NDX daily
data going back to 1986/so:
Probability distribution based on percent decline from top:
0% 10.08
10% 11.33
20% 13.28
30% 14.10
40% 15.51
50% 16.45
60% 19.01
70% 21.84
80% 26.80
90% 38.36
100% 38.36
Probability dist. of a turn, based on number of trading days from the top:
0% 1.00
10% 4.00
20% 5.00
30% 7.00
40% 9.00
50% 12.00
60% 19.00
70% 28.00
80% 49.00
90% 154.00
100% 154.00
The most recent top was on 20020515 with an NDX high of 1350.54.
Friday's low was 1031.95, or a -23.6% drop in 26 trading days.
Referring to the handy tables above, on a percentage drop basis, we're at
about the 75% level ... or there is only a 25% chance, statistically that a
near term bottom with at least a 10% bounce was *not* put in. Our max risk
is that we could go as far as -38.36%, or 832 (yikes) on the NDX (20 on
QQQ). Coincidentally that -38% drop completed on 7/24/2001, after 31
days(!). On a timing basis there's about a 28% chance that the drop has not
yet completed.
Here is the table the data is derived from:
19860703 191 48.83 19851001 51 19.01
19871006 267 60.59 19860916 10 33.48
19871021 1 12.56 19871020 5 21.23
19871106 7 22.08 19871028 19 16.00
19880108 23 30.65 19871204 9 10.39
19880706 115 27.18 19880121 97 15.62
19891010 223 46.85 19881121 77 17.66
19900716 115 25.49 19900130 28 28.27
19900828 3 10.70 19900823 22 13.61
19901002 2 11.65 19900928 8 14.02
19910417 128 75.98 19901012 49 12.68
19920116 141 43.20 19910626 154 19.59
19920914 13 12.88 19920825 15 10.08
19930126 78 31.75 19921005 62 15.11
19940317 227 28.30 19930426 68 16.46
19950717 267 71.42 19940624 2 11.60
19950912 38 15.72 19950719 20 13.99
19951106 19 18.12 19951010 48 15.51
19960606 99 33.65 19960116 27 18.65
19960718 2 12.45 19960716 4 10.29
19970123 127 62.36 19960724 48 16.94
19971009 132 48.09 19970403 13 19.76
19971208 28 17.22 19971028 23 14.04
19980422 69 40.42 19980112 36 11.62
19980721 26 28.34 19980612 11 13.19
19980819 10 10.19 19980805 9 21.34
19980928 18 26.39 19980901 8 24.76
19990201 78 102.28 19981008 12 13.29
19990407 33 20.72 19990218 9 13.28
19990427 5 16.96 19990420 21 14.14
19990719 36 25.90 19990526 16 14.10
19991012 44 21.98 19990810 4 11.07
20000103 53 66.82 19991018 4 13.61
20000124 10 17.81 20000107 5 14.24
20000310 28 39.14 20000131 4 13.09
20000324 6 18.92 20000316 7 26.80
20000410 4 22.22 20000404 4 27.16
20000419 3 19.60 20000414 2 14.72
20000501 5 20.96 20000424 7 16.45
20000516 4 12.95 20000510 6 20.70
20000717 36 41.13 20000524 13 18.27
20000901 21 24.10 20000803 29 27.88
20001017 2 11.20 20001013 1 10.35
20001020 2 17.87 20001018 4 15.90
20001106 7 13.98 20001026 5 18.61
20001115 2 14.07 20001113 10 22.44
20001211 7 23.26 20001130 8 27.29
20001228 4 15.22 20001221 2 15.97
20010104 2 20.97 20010102 2 15.54
20010124 11 28.84 20010108 15 21.58
20010215 1 11.51 20010214 9 24.16
20010306 3 10.39 20010301 12 22.01
20010323 1 10.88 20010322 8 23.15
20010420 11 46.97 20010404 3 12.03
20010502 5 13.58 20010425 10 10.68
20010523 5 17.27 20010516 4 14.71
20010607 6 11.01 20010530 9 15.71
20010629 7 12.69 20010620 7 14.22
20010713 2 11.39 20010711 7 11.33
20010802 7 11.77 20010724 31 38.36
20011026 25 36.59 20010921 2 11.22
20011206 26 31.36 20011030 52 23.33
20020311 11 18.31 20020222 40 27.40
20020515 6 18.24 20020507 21 21.84
The way to read the table is the leftmost 3 columns are date of top, days it
took to get to the top, and the percent change from the recent bottom (low).
The 4th through 6th columns are similarly the date of the bottom, the
duration in trading days of the drop, and the percentage drop from high to
low. The data proceeds in a zigzag sort of fashion.
The probability distribution for *rallies* during the predominately bull
market from 1986 until now are:
probability distribution of percentage change to the upside
0% 10.19
10% 11.51
20% 12.95
30% 17.22
40% 18.92
50% 22.22
60% 28.30
70% 36.59
80% 48.09
90% 102.28
100% 102.28
probability distribution of the length of rally in trading days (without
a -10% or more retracement):
0% 1.00
10% 2.00
20% 4.00
30% 6.00
40% 10.00
50% 19.00
60% 28.00
70% 69.00
80% 128.00
90% 267.00
100% 267.00
thus, a typical rally lasted 19 trading days, and was good for about 22%
bottom to top.
How to use this data? Well, we might consider out of the money one mnnth out
calls on the QQQ. With QQQ at 25.81, a 20% pop would take the Q's to about
31. The July 27 calls are trading for about $1. We might expect, based on
the results above that we've got a decent shot at making $3 on our $1
options, with a 25% chance of losing it all. The expected return is:
0.75 * 3 - 0.25 * 1 = 2.25 - 0.25 = $2.00
This is a really rough estimate, because the analysis doesn't take into
account the full probability distribution of changes. We also need to
consider whether/when we might take profits early.
disclaimer: a lot of money has been lost doing this sort of probability
analysis, because for some reason, the market doesn't listen to this sort of
logic. Still, the bottom line is there actually is a pretty good chance of
seeing a near term bounce - but the downside risk remains substantial if
you're not hedged.
Okay one more table, the distribution of rallies that lasted 20 trading days
or less, without a -10% retracement. First, the percentages:
0% 10.19
10% 11.01
20% 11.65
30% 12.69
40% 13.98
50% 17.27
60% 18.24
70% 20.96
80% 23.26
90% 46.97
100% 46.97
and the duration distribution:
0% 1.00
10% 2.00
20% 2.00
30% 3.00
40% 5.00
50% 6.00
60% 7.00
70% 10.00
80% 11.00
90% 19.00
100% 19.00
so the typical 10% or more bounce (without a 10% or more retracement) lasted
6 days and gained 17%.
------------------------ Yahoo! Groups Sponsor ---------------------~-->
Free $5 Love Reading
Risk Free!
http://us.click.yahoo.com/3PCXaC/PfREAA/Ey.GAA/zMEolB/TM
---------------------------------------------------------------------~->
To unsubscribe from this group, send an email to:
realtraders-unsubscribe@xxxxxxxxxxxxxxx
Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
|