CISCO Futures
1-303-306-1521 1-800 800 7227 Fax 1-303-306-1598
http://www.cisco-futures.com
dljones@cisco-futures.com
Market Waves are NOT Cycles
Copyright CISCO Futures March 25, 2005
A hope of many technical analysts is that markets have a cyclic component
that can be used for predictive purposes. Belief in an underlying periodicity
is implicit in moving average smoothing, since the moving average is
expected to act as a filter. The same is true of oscillators, where two
moving averages of different time frames are subtracted to find differences.
The fact that oscillator differences often move smoothly and change fast
in highly directional markets is taken as evidence that the resulting curves
provide information that can be traded. An underlying but unspoken assumption
of virtually all technical analysis is that market data is well conditioned
(small errors in the data produce small errors in the result). In fact, as
will be shown below, the fundamental assumption of essentially equidistant
peaks in a set of market data is incorrect, and thus market data is poorly
conditioned for most of technical analysis.
It is theoretically
true (Fourier analysis) that several waves
of varying amplitudes and periods can be combined to provide a display that
looks like market flow (with rises, falls and periods of low amplitude).
Unfortunately, it is
also true that market flow can and does come from a complex, self-regulating
process (aggregate decisions of a large number of traders). This process
cannot be described by any known mathematical distribution. Specifically
ruled out is the fourier process, the market is not (so far as anyone can
tell) a combination of sine waves.
Many elements of normal life
are cyclic; the 24 hour day, seasons, weather, ice ages and other natural
phemonena. Some economic phenomena such as booms and busts are thought
to follow some sort of wave, such as the Kondratieff 30 to 60 year wave.
In markets, much has been made of the Elliott five wave up and three
wave down proposition. While phemonena such a Elliott waves become apparent
after the fact, proponents seem unable to agree on the present. Even if you
could tell which leg of an Elliott wave you are in, that knowledge would be
hard to use in day-to-day trading where your outlook is rarely more than
just a few minutes or days.
Markets always display wave-like motion. Prices move up and down with
some regularity, describing local highs and lows. This movement reflects the
changes in the way the composite of all traders, their feedback, value
the market. While clearer on longer term
charts, the local highs and lows describe periods that vary from wave to wave.
Prediction of the next high or low cannot be made in any reliable way. Incidentally,
prediction is not a requirement of successful trading (see the section on
the Market Unit).
The Market Unit
Because the source of market motion cannot be tied to any known distribution,
a market surely cannot be described by a set of sine waves or helped in any
real analytical sense by smoothing.
Some analysts describe market waves as cycles, imputing a regularity not supported
by the data itself. Analyzing market waves to obtain any predictability
is not simple and may well be impossible,
due to wave irregularity. Some waves can be very short, comprising just
a day or two from top to bottom, others may take twenty days to find a bottom.
While market waves for predicting the
future seem to be impossible to divine, it is possible to look to the
past to study the obvious wave motion in past data, with the aim of better
understanding markets. We show in figure 1 below how widely the SP market
has varied through the years. Figure 1 shows that the wave ranges of the
SP have been decreasing since 2002 and helps explain why it is harder to
trade the SP profitably today, as compared to the past.
We identify the historical wave motion with a 'seven day' rule--a peak is
defined as the highest
price reached since the previous low with no higher price within the
past seven days. The low is generally defined the same way, but in cases
of disagreement, say with a very short wave, the procedure is modified.
In the two short tables below, the SP futures of March 2000 and 2005 show a
remarkable variability in period (time from peak to peak) and amplitude (range from
peak to trough) is evident. The over-heated markets of 2000 had an average
period of 14.5 days (with large variations) and an average amplitude of
$25,000. The quieter time of 2005 displays a 25% longer period and an
average amplitude 68% less.
SP MAR 00 DATE PEAK VALLEY CAL DY MKT DY Period Ampl 11/16/991450.00 175 121 12 12/ 1/99 1406.50 190 131 43.50 12/ 3/991470.00 192 133 20 12/ 9/99 1410.50 198 137 59.50 1/ 3/ 01496.50 223 153 9 1/ 5/ 0 1387.00 225 155 109.50 1/14/ 01484.50 234 162 17 1/31/ 0 1357.00 251 172 127.50 2/ 9/ 01454.80 260 179 2/28/ 0 1327.00 279 191 127.80Wave periods and amplitudes for the first quarter of 2000. This study at the height of the telecommunications bubble shows the cross market effect on the SP index in the wave amplitudes. A 100.0 range is $25,000. Wave times (periods) averaged 14.5 days. The average range per day is about $1,700; indicating much opportunity for the day trader.
SP MAR 05 DATE PEAK VALLEY CAL DY MKT DY Period Ampl 11/17/ 41191.50 597 335 30 11/22/ 4 1171.00 602 338 20.50 12/31/ 41219.70 641 365 13 1/19/ 5 1176.70 660 378 43.00 1/19/ 51197.20 660 378 20 1/24/ 5 1166.10 665 381 31.10 2/16/ 51213.40 688 398 13 2/22/ 5 1184.00 694 402 29.40 3/ 7/ 51225.70 707 411 3/16/ 5 1185.60 716 418 40.10Wave periods and amplitudes for the first quarter of 2005. Five years after 2000, in a much cooler market, wave periods average 18.25 days and amplitudes are around 31 (approximately $8,000). Daily opportunity averages about $450. Clearly, from the standpoint of opportunity alone, it is harder to make money trading the SP in 2005.
Average Wave Amplitudes for SP 1990 - 2005
Ampl
100 - + xxx xxxxx xx
x x x
90 - x
x x
80 -
x x x
70 - x
x x
60 -
x x
50 - x x
x x
40 - x x x
x x
30 - x xx
xxxx x
20 -
xxxxxxxxx xxxx
10 - xxx xxx
0 -
| | | | | | | | | | | | | | | | |
90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06
Years
Figure 1. Average trading wave amplitude for SP. Quarterly averages
(four per year), provide the average trading opportunity (price range
from high to low in a wave). Wave peaks are the highest prices within
the previous 7 trading days. The average number of days in a wave,
calculated quarterly (not shown on this graphic) varies from a low of 11
(second quarter of 1998) to a high of 43 (fourth quarter of 1998). Wave
amplitudes (price range from high to low in a wave, averaged per quarter)
were as small as 11.2 (4th Qtr 1993 and 1st Qtr 1995) and as large as
155.5 ($38,875) )(4th Qtr 2000). Ranges above 100 are truncated on this graph.
Del NRng Rng-Hi Rng-Lo Rng-Av NPd Pd-Hi Pd-Lo Pd-Av Pri
DEC 89 6 40.35 4.40 15.86 6 21 7 14 360
MAR 90 2 44.20 15.80 30.00 2 27 26 26 350
JUN 90 2 22.20 12.85 17.53 2 35 18 26 350
SEP 90 4 40.15 19.20 28.86 4 28 16 21 350
DEC 90 3 27.90 12.35 19.57 4 17 8 13 320
MAR 91 3 25.40 9.80 15.47 3 42 8 20 340
JUN 91 4 19.30 12.25 16.54 4 20 9 15 390
SEP 91 4 23.90 7.90 16.71 4 32 12 19 385
DEC 91 4 28.10 6.15 15.38 4 21 9 15 385
MAR 92 4 15.20 11.05 13.40 3 20 11 16 410
JUN 92 3 20.30 9.25 13.68 3 17 13 15 410
SEP 92 3 19.00 8.80 14.87 3 31 14 24 410
DEC 92 2 35.50 7.10 21.30 2 35 25 30 420
MAR 93 3 22.30 6.25 15.08 3 28 14 21 430
JUN 93 4 18.85 10.00 14.00 4 18 8 14 445
SEP 93 4 13.60 7.80 10.75 4 35 7 18 450
DEC 93 4 17.15 7.05 12.48 4 22 8 15 460
MAR 94 5 17.50 3.50 9.95 5 23 8 15 475
JUN 94 3 36.30 13.30 20.98 3 31 11 19 450
SEP 94 4 24.65 5.40 12.88 4 22 11 16 460
DEC 94 4 32.40 14.40 20.21 4 21 8 13 460
MAR 95 3 13.80 7.40 10.93 3 40 18 27 465
JUN 95 2 13.75 9.35 11.55 2 33 13 23 510
SEP 95 5 21.55 9.60 13.91 5 21 12 15 560
DEC 95 3 21.35 13.05 17.17 3 15 7 12 580
MAR 96 3 42.90 18.65 31.32 3 28 13 21 625
JUN 96 2 39.50 27.90 33.70 2 27 19 23 645
SEP 96 3 71.50 24.70 41.97 3 41 26 34 670
DEC 96 1 20.10 20.10 20.10 1 26 26 26 710
MAR 97 4 50.00 33.00 38.23 4 22 14 17 760
JUN 97 2 69.30 41.00 55.15 2 43 10 26 790
SEP 97 3 76.05 22.75 43.52 3 33 9 20 890
DEC 97 5 148.25 26.70 63.97 5 22 11 15 940
MAR 98 1 77.50 77.50 77.50 1 14 14 14 955
JUN 98 5 74.50 37.20 52.78 4 18 8 11 1105
SEP 98 3 176.50 56.70 125.33 3 27 22 24 1130
DEC 98 1 151.40 151.40 151.40 1 43 43 43 1120
MAR 99 3 89.70 61.00 75.23 2 16 14 15 1220
JUN 99 5 104.00 61.10 77.52 5 17 11 14 1340
SEP 99 3 158.00 49.30 96.27 3 29 8 21 1375
DEC 99 3 112.00 46.90 89.33 3 26 12 20 1400
MAR 00 4 126.00 41.30 97.20 4 32 9 19 1440
JUN 00 4 201.10 112.50 142.03 4 14 11 12 1440
SEP 00 3 107.20 41.80 70.33 3 34 11 21 1515
DEC 00 3 157.80 119.50 144.47 3 29 18 23 1440
MAR 01 3 283.40 77.50 166.47 3 38 16 24 1340
JUN 01 2 83.10 28.50 55.80 2 24 3 13 1260
SEP 01 4 76.40 63.00 70.60 4 28 10 16 1230
DEC 01 3 256.50 26.30 111.13 3 30 9 22 1080
MAR 02 5 103.00 6.00 49.08 5 27 7 14 1130
JUN 02 3 130.10 76.70 98.90 3 22 20 21 1080
SEP 02 2 271.90 96.10 184.00 2 46 13 29 930
DEC 02 2 159.50 56.00 107.75 2 39 17 28 900
MAR 03 3 129.00 34.70 83.00 3 28 9 20 840
JUN 03 2 63.50 51.80 57.65 2 14 9 11 840
SEP 03 6 55.10 27.10 39.07 5 47 13 25 980
DEC 03 3 50.10 32.90 40.57 3 21 12 16 1030
MAR 04 2 33.50 24.00 28.75 2 15 13 14 1130
JUN 04 2 74.50 36.30 55.40 3 23 11 17 1130
SEP 04 4 67.50 18.90 48.95 4 33 12 24 1090
DEC 04 2 55.60 26.00 40.80 2 30 13 21 1100
MAR 05 4 43.00 20.50 31.00 3 30 13 21 1200
Headings:
Del = delivery
NRng = number of waves per quarter
Rng-Hi = the largest high-low range per quarter
Rng-Lo = is the smallest high-low range per quarter
Rng-Av = average high-low range
NPd = the number of waves in a calculation
Pd-Hi = the largest number of days in a wave
Pd-Lo = the least number of days in a wave
Pd-Av = the arithmetic average number of days in a wave.
Pri = approximate average price for the quarter
Grouping the period lengths shows their general distribution:
# Days Pd-Hi Pd-Lo
40 - 49 7 1
30 - 39 16 0
20 - 29 28 11
10 - 19 11 30
3 - 9 0 20
Historically, from the time of Welles Wilder's book on technical analysis
in 1978, the choice for averaging
was fourteen days. For the low periods, that falls in about the middle.
However, the median for the high periods is about ten days more or around
twenty five days. Again, there is not even a vestige of a median to average
around. The consequence is that averaging, as in moving averages and
oscillators, is close to dealing with random numbers.
Crude Oil
Del NRng Rng-Hi Rng-Lo Rng-Av NPd Pd-Hi Pd-Lo Pd-Av Pri
MAR 90 5 2.28 0.38 0.99 5 29 9 17 19
JUN 90 3 3.55 0.80 2.15 3 32 8 18 21
SEP 90 3 2.65 0.92 1.73 3 18 7 12 19
DEC 90 4 4.91 0.85 3.17 4 30 8 19 30
MAR 91 6 12.20 2.90 6.79 6 34 5 14 31
JUN 91 3 2.40 0.97 1.76 3 22 11 16 20
SEP 91 4 1.83 0.72 1.11 4 41 8 18 21
DEC 91 3 1.42 0.62 0.96 3 35 9 19 22
MAR 92 3 4.70 0.73 2.47 3 40 9 20 22
JUN 92 2 1.53 0.66 1.10 2 41 17 29 20
SEP 92 3 1.77 0.64 1.03 3 30 21 25 21
DEC 92 2 1.10 0.61 0.86 2 33 24 29 22
MAR 93 4 1.80 0.69 1.49 4 23 7 19 20
JUN 93 3 1.40 0.89 1.21 3 25 21 22 21
SEP 93 2 3.45 1.02 2.24 2 40 19 30 20
DEC 93 3 2.06 0.48 1.11 3 24 11 16 19
MAR 94 3 3.60 1.26 2.31 3 32 20 26 17
JUN 94 2 1.71 0.96 1.34 2 18 14 16 15
SEP 94 4 1.10 0.92 1.02 4 34 8 19 18
DEC 94 4 1.89 0.72 1.65 4 28 14 21 19
MAR 95 3 1.58 0.70 1.15 3 27 13 18 18
JUN 95 2 0.56 0.45 0.51 2 47 20 34 19
SEP 95 3 2.31 0.60 1.37 3 30 4 16 19
DEC 95 5 1.28 0.43 0.75 5 22 8 14 17
MAR 96 3 2.49 0.32 1.07 3 43 8 31 17
JUN 96 3 2.68 1.18 1.85 3 35 9 25 20
SEP 96 3 1.81 1.06 1.43 3 20 17 19 20
DEC 96 3 1.88 0.80 1.29 3 28 17 23 22
MAR 97 4 2.42 1.13 2.01 4 22 11 17 24
JUN 97 2 3.65 3.07 3.36 2 33 17 25 22
SEP 97 3 3.20 0.73 1.96 3 34 8 18 21
DEC 97 3 2.35 0.69 1.52 3 22 10 17 21
MAR 98 2 5.41 0.90 3.16 2 50 14 32 21
JUN 98 2 4.86 2.10 3.40 2 36 18 27 17
SEP 98 5 2.56 1.06 1.73 5 23 10 14 16
DEC 98 4 3.07 0.76 1.55 4 22 11 17 15
MAR 99 3 3.66 1.86 2.52 3 31 14 20 14
JUN 99 2 1.35 1.32 1.34 2 41 25 33 16
SEP 99 3 1.54 0.28 0.95 3 22 10 17 18
DEC 99 2 4.18 1.44 2.81 2 26 18 22 23
MAR 00 3 2.55 1.59 2.16 3 23 15 19 25
JUN 00 1 1.35 1.35 1.35 1 32 32 32 28
SEP 00 5 5.21 1.35 3.90 5 23 10 16 28
DEC 00 5 4.56 1.53 2.90 5 40 10 32 33
MAR 01 3 8.39 0.44 3.29 3 33 5 18 31
JUN 01 3 2.86 1.94 2.39 3 17 13 15 28
SEP 01 5 4.00 0.42 2.83 5 16 5 11 29
DEC 01 5 8.10 0.36 2.46 5 20 8 13 28
MAR 02 3 3.84 2.14 3.23 3 20 15 18 22
JUN 02 2 2.75 1.16 1.96 2 32 26 29 22
SEP 02 4 4.40 1.55 2.65 4 40 9 19 27
DEC 02 4 1.30 0.64 1.02 4 22 4 13 28
MAR 03 1 4.36 4.36 4.36 1 53 53 53 30
JUN 03 4 8.93 2.37 4.43 4 32 12 18 30
SEP 03 3 2.25 2.20 2.22 3 34 20 25 31
DEC 03 2 4.63 1.59 3.11 2 34 10 22 31
MAR 04 3 3.00 2.80 2.92 3 26 15 20 32
JUN 04 3 4.78 1.85 3.23 3 44 4 22 34
SEP 04 4 5.50 1.55 3.11 4 30 11 21 42
DEC 04 2 10.65 6.25 8.45 2 45 19 32 51
MAR 05 2 9.40 5.55 7.48 2 26 13 20 48
JUN 05* 2 3.85 2.91 3.38 2 30 6 18 (55)
incomplete
Headings:
Del = delivery
NRng = number of waves per quarter
Rng-Hi = the largest high-low range per quarter
Rng-Lo = is the smallest high-low range per quarter
Rng-Av = average high-low range
NPd = the number of waves in a calculation
Pd-Hi = the largest number of days in a wave
Pd-Lo = the least number of days in a wave
Pd-Av = the arithmetic average number of days in a wave.
Pri = estimated average price for the quarter
The timeframe MAR 90 through Jun 00 is one of low opportunity and
low trading opportunity (a range of 1.0 is $1000). Average wave length
for this period is 22 days. From Sep 00 through the present, average
wave amplitudes are about double of the earlier ones. Average wave length
for the later period is 22 days, same as for the low amplitude region.
What the market is doing, is simply offering more opportunity in the
same elapsed time.
Data from Advice Engine Select Tables: Trading Opportunity for Crude Oil 2000 - 2005
Crude Oil Annualized Opportunity Averages Per Trade
$ |
|
650 |
| X
600 | X
|
550 | X X
|
500 |
|
450 |
|
400 | X
| X
350 |
|
300 | X
|
250 | X
01 01 02 02 03 03 04 04 05 05
Years