FR H-Volatility - page 39

New comment
 

Mathemat, hi!

You have to shrink the pictures - they don't fit in the monitor. I have to move the mouse back and forth! Of course, I understand that they are very interesting and you tried to convey ALL the information, but, man, not to the same extent:-)

 
Neutron:

I think the problem is, more than you pointed out, the width of the distribution.


The problem of width of distribution is mitigated by abandoning the objective of "working at full stroke of all segments".
 
You've squeezed it, Neutron. I've had it before... But if the intellectual community specifically requests it, that's no problem. What resolution does your monitor have?
 

Ehhh, tried it on the clocks and this is what came out (top are astronomical bars, bottom are equivolume bars):

Don't pay attention to the numbers of bars, it's not that important. What is important is that the pictures actually contain the same stories - from December 6 till December 24, 2007.

 
Mathemat:

Ehhh, tried it on the clocks and this is what came out (top are astronomical bars, bottom are equivolume bars):

Don't pay attention to the number of bars, it's not that important. What is important is that the pictures actually contain the same stories - from December 6 till December 24, 2007.


You argue that the number of bars is not important, but IMHO if you scale the x-axis, you get an advance/delay of equi-bars relative to astronomical bars. Maybe there will be some patterns? Or the anticipation/delay could be considered as a trend/flat?
 
Or could the lead/lag be seen as a trend/flat?

Yes, that's exactly what I was counting on. It's a matter of figuring out how to scale the X axis.

 
lna01:
Neutron:

I think that the problem is, to a greater extent than you pointed out, in the width of the distribution.


The problem of spreading width is mitigated by abandoning the aim of 'working at full stroke for all segments'.


It is the constancy of the width of distribution (and its all-importance) at all scales of the ZZ that has led me to fail to find a way to use the very dependence discussed above.

However, I did draw one conclusion for myself: since there is such a statistical dependence, there is no point in poking around in the duration of the ZZ segments, it is enough to examine the size of the ZZ.

 
Yurixx:


It is the constancy of the distribution width (and its all-importance) at all scales of WP that has led me to fail to find a way to use the very dependence I spoke about above.

However, I have made one conclusion for myself: since there is such a statistical dependence, there is no point in poking around in the duration of the SPZ segments, it is enough to examine the size of the SPZ.

Neither have I yet been able to find a practical way to use the "statistical targets" approach. But the topic continues to seem interesting. I use the time information, it allows me to add one more characteristic (besides the size of the gZ, which by the way is also used in a not quite direct way).
 
Mathemat:
Or could the lead/lag be seen as a trend/flat?

Yes, that's exactly what I was counting on. I still need to figure out how to scale the X-axis.


And this is what I got. Blue minutes from Alpari, red ticks from here http://ratedata.gaincapital.com/ (first week of December GBPUSD)

The beginning and the end correspond to time, I spent half a day trying to figure it out. I've been looking at this picture all night, trying to figure out where I screwed up.

 
Mathemat:
Or could the lead/lag be seen as a trend/flat?

Yes, that's exactly what I was counting on. It's a matter of figuring out how to scale the X axis.


What if we simply calculate Bars*TpB - Ticks, TpB is the average number of ticks per bar? More precisely the derivative of this value, to get rid of uncertainty with the datum. I mean to do without constructing equi-bars.
1...3233343536373839404142
New comment