Market phenomena - page 15

 
Avals:

not to the holes - more subtle structures are at work :)
They will come to nanotechnology. They will start nanopiping, the Galileans. They will have to book seats at Skolkovo with Dmitry Anatolyevich :))
 
paukas:
That's how they'll get to nanotechnology. They are going to start nanoprescribing. They will have to book seats at Skolkovo with Dmitry Anatolyevich:))

Do you know what is going on in Skolkovo? There will be plenty of room for you...
 
Cmu4:

Do you know what's going on in Skolkovo? You'll have plenty of room there.
Of course I do. Just like everywhere else. Stealing.
 
Avals:

Not really :) if the first column represents a sequence of gyrations, and the second one represents how many increments "got" into it, then what will happen when the interval size is less than a point? For example, let EURUSD m15 change from +100 to -100 pips. I.e. the range = 200 points. Let us take 700 intervals. The width of one of them will be 200/700=0.29 points. In reality, the increments are a multiple of a whole number of points. For example, there will be 3 empty intervals between increments of 5 points and 6 points where nothing should fall, unless you move to an additional sign later ;) I.e. the number of intervals should not exceed the range in points

Yes, moreover, according to my understanding, the number of intervals should not exceed the value of spread/2 (the only exception is 1/1 ratio). I think a reference to Kotelnikov's theorem would be appropriate.

One more thing. In this case it is highly desirable, even if the above condition is observed, that the discreteness of measured intervals be strictly a multiple of the point size. Otherwise we will come across the same "phenomenon", as in some intervals there will be more measured points than in others. Especially with "highly rational" ratios (c) the "phenomenon" will be more pronounced with a 5/3 or 7/4, than with 37/29 or 23/16.

Unfortunately researchers often buy into phenomena that have their roots only in peculiarities of information processing technology.

 
MetaDriver:

Yes, moreover, according to my understanding, the number of intervals should not exceed the value of spread/2 (the only exception is 1/1 ratio). I think a reference to Kotelnikov's theorem would be appropriate.

One more thing. In this case it is highly desirable, even if the above condition is observed, that the discreteness of measured intervals be strictly a multiple of the point size. Otherwise we will come across the same "phenomenon", as in some intervals there will be more measured points than in others. Especially with "highly rational" ratios (c) the"phenomenon" will be more pronounced, i.e. with 5/3 or 7/4, than with 37/29 or 23/16.

Unfortunately, researchers often buy into phenomena that have their roots only in peculiarities of information processing technology.


Yes, the spread should be a multiple of the number of intervals. That's why it's more convenient to take an interval width of 1 point, so you don't have to make any fuss. It's easy and straightforward :)
 

Here's another phenomenon, or maybe a freak :)

We plot the distribution of the distance between neighbouring highs of bars, or loys.
I.e. we take High[i]-High[i+1] and look at the distribution.

Comparing this distribution of eurusd m15 and the random walk generated on the basis of its volatility (sample volumes are equal). On the abscissa axis is the distance in pips from the high of the previous bar. I.e. 0 is the level of the high of the previous bar:

the real distribution is more peaked, oblique and "tailed". The right part of the peak is actually where the price stopped more often after breaking the extremum of the previous bar.

The interesting ones are the small peaks of exactly +20p and -20p. They are not very visible because for the m15 bar it is an extreme distance. Rarely does an m15 bar get there)))

Here are the distributions on h4 for several majors:

levels of +20p and -20p are shown in red, +40p and -40p are shown in blue

I.e. the probability of the high bar being exactly 20 points above or below the high of the previous bar is significantly different from a random one. It appears even more strongly at higher frames (though there are less statistics there)

In general, 20 pips old is a large enough value to be a rounding error, sampling, etc. Maybe there really are more limiters set there and these levels become support/resistance more often. Or maybe this is peculiarities of broker's rounding of historical data or something :) But if so, maybe when trading a breakdown of an extremum it is better to close before that peak, i.e. TP is somewhere 17-18 pips behind the breakdown level.

 
Avals:

Here's another phenomenon, or maybe a freak :)


Not a freak and not a phenomenon. I cannot say for the given data, but in other places this phenomenon has been observed. And traders (based on literary sources), with a pita or close to it, report about this phenomenon. Stops are very predictably set by the crowd and institutions on some symbols.

On currencies, it's said that London likes to go after such stops.

In general, the effect seems to take place, but not everywhere and not all the time. It is the same as always, nothing constant in the market.

 
MetaDriver:

Yes, moreover, according to my understanding, the number of intervals should not exceed the value of spread/2 (the only exception is 1/1 ratio). I think a reference to Kotelnikov's theorem would be appropriate.

One more thing. In this case it is highly desirable, even if the above condition is observed, that the discreteness of measured intervals be strictly a multiple of the point size. Otherwise we will come across the same "phenomenon", as in some intervals there will fall more measured points, than in others. Especially with "highly rational" ratios (c) the "phenomenon" will be more pronounced with a 5/3 or 7/4, than with 37/29 or 23/16.

Unfortunately, researchers often buy into phenomena that have their roots only in peculiarities of information processing technology.


It seems that you and Avals have only read the first few sentences. I'll wait until you've mastered the whole text. I'm in no hurry.

 
paukas:
Explain to me, country boy. Are there any holes, or have they run out?

You have holes, not filled with anything yet.
 
Colleagues, once again - the non-standard histogram only suggested the existence of a classification (filtering) which divides the quote stream into two opposite sub-flows with very "smooth" characteristics. That's all, nothing more. Leave this histogram alone.(I'll hide the details, sorry).
Shit... no, that's the wrong term, &&&& - there, that's more correct.