Analysis and Trading based on Envelopes, Waves, Cycles | Page 5 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Planetary cycles, astrological theory of cycles of the markets 510 replies

Forecast Forex using Numeric Cycles and Astrology Cycles 38 replies

Planetary Cycles... Plotting Cycles into the Future 0 replies

Market Waves are NOT Cycles 8 replies

Waves/Trends/Cycles 0 replies

Attachments: Analysis and Trading based on Envelopes, Waves, Cycles
Exit Attachments

Analysis and Trading based on Envelopes, Waves, Cycles

  • #81
  • Oct 27, 2017 4:31am Oct 27, 2017 4:31am
WTI - 6 Months
Attached Image (click to enlarge)
Click to Enlarge Name: OIL WTI 6 MONTHS 27 OCTOBRE 2017.jpg Size: 219 KB
  • #82
  • Edited 2:07pm Oct 28, 2017 1:56pm | Edited 2:07pm
Just recalled that I had this code in ProRealTime.

Here's a 15 minute chart of the S&P 500 from the middle of October with a 60 period centered MA (red) and a channel based on an octic filter (yellow).

Should anybody be interested, here's the ProRealTime code:

Inserted Code
ONCE p2 = round(0.6*p1)
ONCE s2 = p1*(p1-1)*(2*p1-1)/6
ONCE s4 = p1* (p1-1)*(2*p1-1)*(3*p1*p1-3*p1-1)/30
ONCE sw = p1 + s4/(p1*p1*p2*p2) -s2*(1/(p1*p1)+1/(p2*p2))
 
if  barindex < p1 then
qf = undefined
else
qf = 0
FOR i = 0 To p1 - 1
ww = (1-i/p1)*(1+i/p1)*(1-i/p2)*(1+i/p2)
qf=  qf + ww* close[i]
NEXT
endif
qf = qf/sw
 
uc = qf + d
lc = qf - d
return qf, uc, lc

p1 and d are the variables. p1 is 120 in the charts, so corresponds to twice the "normal" period, and the channel distance is 3 points.
Attached Image(s) (click to enlarge)
Click to Enlarge Name: Screen Shot 2017-10-28 at 20.48.09.png Size: 153 KB
Click to Enlarge Name: Screen Shot 2017-10-28 at 20.51.29.png Size: 142 KB
  • #83
  • Oct 28, 2017 3:01pm Oct 28, 2017 3:01pm
SP 500 - 24 H / 1 minute
Attached Image (click to enlarge)
Click to Enlarge Name: SP500 11 28 OCTOBRE 2017.jpg Size: 157 KB
  • #84
  • Oct 28, 2017 3:02pm Oct 28, 2017 3:02pm
SP 500 - 5 Jours / 5 minutes
Attached Image (click to enlarge)
Click to Enlarge Name: SP500 55 28 OCTOBRE 2017.jpg Size: 184 KB
  • #85
  • Oct 28, 2017 3:02pm Oct 28, 2017 3:02pm
SP 500 - 1 Month
Attached Image (click to enlarge)
Click to Enlarge Name: SP500 1 MONTH 28 OCTOBRE 2017.jpg Size: 171 KB
  • #86
  • Oct 28, 2017 3:03pm Oct 28, 2017 3:03pm
SP 500 - 3 Months
Attached Image (click to enlarge)
Click to Enlarge Name: SP500 3 MONTHS 28 OCTOBRE 2017.jpg Size: 180 KB
  • #87
  • Oct 28, 2017 3:04pm Oct 28, 2017 3:04pm
SP 500 - 6 Months
Attached Image (click to enlarge)
Click to Enlarge Name: SP500 6 MONTHS 28 OCTOBRE 2017.jpg Size: 187 KB
  • #88
  • Edited 5:54pm Oct 28, 2017 4:58pm | Edited 5:54pm
Quoting parisboy
SP 500 - 6 Months {image}
Here is the same with the octic filter channel for comparison. And reflecting on that chart, and the bull flags added to it, one can see how the channel cycle analysis neatly captures the structure of the trend.
Attached Image (click to enlarge)
Click to Enlarge Name: Screen Shot 2017-10-28 at 23.54.25.png Size: 314 KB
  • #89
  • Oct 28, 2017 5:30pm Oct 28, 2017 5:30pm
Linear regression channels work well too. This is the Dow Jones Industrial Average.
Attached Image (click to enlarge)
Click to Enlarge Name: Screen Shot 2017-10-29 at 00.29.12.png Size: 195 KB
  • #90
  • Oct 28, 2017 5:44pm Oct 28, 2017 5:44pm
Quoting medici
{quote} Here is the same with the octic filter channel for comparison. And reflecting on that chart, and the bull flags added to it, shows how the channel cycle analysis neatly captures the structure of the trend. {image}
Sorry Medici , what is the "octic filter channel" ?
  • #91
  • Oct 28, 2017 5:48pm Oct 28, 2017 5:48pm
Quoting parisboy
{quote} Sorry Medici , what is the "octic filter channel" ?
It's produced by the code I posted earlier. It is based on digital filtering techniques, i.e. the kind used, for example, to filter out high-frequency noise from signals. Such filters tend to lie in the middle of the price series, just like the centered MA.

The code is from years ago when I had access to institution grade computational platforms and played with these things a bit.
  • #92
  • Oct 28, 2017 5:56pm Oct 28, 2017 5:56pm
Medici thank you for your answer

because I found no information via google
  • #93
  • Oct 28, 2017 6:22pm Oct 28, 2017 6:22pm
Quoting parisboy
Medici thank you for your answer because I found no information via google
If you are interested, "octic" refers to 8th degree polynomials, and the channel is based on octic polynomial regression.

In public forums, you will find plenty of debate about polynomial regression being "repainting", i.e. the regression line as a whole gets updated as new data points come in, which clearly isn't backtestable for trading purposes.

The above filter does not repaint. It is a simple version of what is described in this article:

https://en.wikipedia.org/wiki/Savitz...93Golay_filter
  • #94
  • Oct 29, 2017 2:43am Oct 29, 2017 2:43am
Thank you Medici

it is always stimulating for the mind to read such things !

But basically I am (almost) illiterate in such mathematics and much more in implementing them on platforms for trading !
  • #95
  • Oct 29, 2017 7:56am Oct 29, 2017 7:56am
Quoting parisboy
Thank you Medici it is always stimulating for the mind to read such things ! But basically I am (almost) illiterate in such mathematics and much more in implementing them on platforms for trading !
You're welcome Having done enough of such things in the past it was a kind of déjà vu to recall that filter, as it's several years ago that I looked at it.

And I am grateful to you for bringing Hurst cycle analysis to my attention Having thought about it, it seems like the multifractal scaling of periods and channel distances is the essential thing. For practical trading, I think I prefer linear regression channels to both centered moving averages and more mathematical constructs.

Here's the DAX from late 2015 and this autumn. Some similarities....

It's a 2 hour chart with regression channels (192, 4), (768, 8), and (1536, 16).
Attached Image(s) (click to enlarge)
Click to Enlarge Name: Screen Shot 2017-10-29 at 13.53.34.png Size: 303 KB
Click to Enlarge Name: Screen Shot 2017-10-29 at 13.53.48.png Size: 236 KB
  • #96
  • Oct 30, 2017 5:36am Oct 30, 2017 5:36am
Attached Image (click to enlarge)
Click to Enlarge Name: DAX 11 30 0CTOBRE 2017.jpg Size: 178 KB
  • #97
  • Oct 30, 2017 7:02am Oct 30, 2017 7:02am
Attached Image (click to enlarge)
Click to Enlarge Name: DAX 11 30 0CTOBRE 2017.jpg Size: 172 KB


See the Envelopes parameters circled in red

- for the Envelope 256 Time Unit = 31,25
- for the Envelope 128 Time Unit = 15,625
- for the Envelope 32 Time Unit = 7,8125

let's assume that presently DAX price fluctuates between 13.000 and 13.250

13.250 - 13.000 = 250 points

250 / 8 = 31,25
  • #98
  • Oct 30, 2017 7:09am Oct 30, 2017 7:09am
please ask all technical questions not by private message but publicly on the thread so that everybody could benefit from the answer
  • #99
  • Oct 30, 2017 7:48am Oct 30, 2017 7:48am
seen on another forum
Attached Image (click to enlarge)
Click to Enlarge Name: eurousd 4 H IB 30 OCTOBRE 2017.jpg Size: 102 KB
  • #100
  • Oct 30, 2017 7:49am Oct 30, 2017 7:49am
Attached Image (click to enlarge)
Click to Enlarge Name: EURUSD 11 30 OCTOBRE 2017.jpg Size: 171 KB
10 traders viewing now