Market phenomena - page 12
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
hmmm... now I notice that somehow it worked out.... :D))))
but by golly, unintentionally....
The phenomenon of long-term memory has another "angle" - it is the long tails of the distribution itself. Roughly speaking (not even "near-scientific"), all memory sits in these very long tails. I.e. the occurrence of (some given) evasion event (there are subtleties in calculation) is significantly increased for a process with long tails of increments (as opposed to).
PS: the trajectory bias is the sum of the increments on a plot of a given length
Thick tails, islanding etc. appear perfectly well in ARCH/GARCH distributions that have no directional memory - only volatility. I.e. the presence of fat tails does not confirm the presence of a long-term memory that can be used for making profit (based on the difference in the price of an asset at different points of time)
I forgot to add to my post:
у феномена долговременной памяти есть и другой "ракурс" - это сами длинные хвосты распределения. Если очень грубо (даже не "околонаучно") - вся память в этих самых длинных хвостах сидит. Т.е. существенно повышается появление события (какого то заданного) уклонения (есть тонкости в подсчете) у процесса с длинными хвостами приращений (в отличии от).
Thesame effect of highlighting the alpha and omega processes can be obtained by dividing the quoted incremental flow into, what is "less lambda" and what is "more lambda". Where lambda is an estimate of the boundary of the long tail of the incremental distributions. The exact lambda estimate is hard to find, but that's how I first got the two processes (for M15 EURUSD L would be about 2-3 RMS):
PS: Try it if anyone is interested (I just won't post it until Sunday at the earliest). Impressive to see :o)
Sergey, do you have specific studies with confirmations (preferably on mql4)?
yes, but on MathCAD, and here:https://www.mql5.com/ru/forum/134424/page12 is a continuation of the phenomenon. The described phenomenon and was found on the basis of long term memory analysis (a long time, my favourite subject) and tedious research with frequencies.
The thick tails, isovershinness, etc. show up perfectly in ARCH/GARCH distributions, which have no directional memory - only volatility.
To begin with, ARCH/GARCH do not allow for a qualitative simulation of "strictly" quotient processes. Normal distributions, t-distributions, I think there is a variant for the conditional, and a few more with complex identification of the model itself, mostly by the likelihood method. Perhaps there is - then please give me the source. But let me remind you that when the forecast model distribution does not coincide with the distribution of the source, one cannot make a quality forecast (it is fundamentally impossible), which is what the ARCH/GARCH models demonstrate at the real quotes. Unfortunately, they can't. If not - just show me how, it would be interesting. I know the acronyms no less :o)
I.e. the presence of thick tails does not confirm the long-term memory on which one can make profit (on the difference of asset prices at different moments)
I didn't look at the tails, I used fractal analysis for my studies. And the influence of tails is interesting in itself they have a huge influence on evasions, huge.
To begin with, ARCH/GARCH do not allow for a qualitative simulation of "strictly" quotient processes. Normal distributions, t-distributions, I think there is a variant for the conditional, and a few more with complex identification of the model itself, mostly by the likelihood method. Perhaps there is - then please give me the source. But let me remind you that when the forecast model distribution does not coincide with the distribution of the source, one cannot make a quality forecast (it is fundamentally impossible), which is what the ARCH/GARCH models demonstrate at the real quotes. Unfortunately, they cannot. If not - just show me how, it would be interesting. I know just as much about acronyms :o)
What can I give you? How do you generate GARCH? It does not matter how well ARCH/GARCH or other implementations describe real quotes (you can ask google ;)), the point of the post was that the presence of thick tails does not indicate long or short term memory, but there is for example an elementary manifestation of autocorrelation and volatility clustering. If you practically want to get a sb close to the real distribution - above is a script that generates a sb based on the volatility of a real instrument. Thick tails, spikiness, etc. like real quotes. Or do you mean something else by "qualitatively simulate "strictly" quotation processes"?
So your statement "all memory sits in these very long tails" is incomprehensibly based on what. How did you arrive at it? If on this:
I didn't look at the tails either, I used fractal analysis to study it. And the influence of tails is interesting in itself they have a huge influence on evasions, huge.
it's not clear what and how you analyzed and what the evasions were
P.S. as for the phenomenon on the first page - double-checked and didn't get it. Have you tried checking on quotes from other DCs and/or other software?
what to give you? how to generate GARCH? It doesn't matter how well ARCH/GARCH or other implementations describe real quotes (you can ask google that ;)),
You'd better ask google first. That's if serious and without :o)
The essence of the post was that the presence of thick tails does not imply long or short term memory, but there is for example an elementary manifestation of autocorrelation and volatility clustering.
You had two main points in your post.
Answered you on each.
If you want a near real distribution - above is a script that generates a sb based on the volatility of a real instrument. Thick tails, spikiness, etc. like real quotes. Or do you mean something else by "qualitatively simulate "strictly" quotation processes"?
I asked clearly and concisely - about your practice of using ARCH/GARCH. To make it clearer - it's unlikely that you or anyone else will sustainably earn money using only ARCH/GARCH. These methods have no practical advantages. Unfortunately. Don't misunderstand (in literary terms) if such "monsters" (this one in particular - Volterry series):
don't "take" quotes and volatility, then the primitive ARCH/GARCH all the more can't handle it ....
So your statement "all memory sits in these very long tails" is incomprehensibly based on. How did you arrive at it?
I wrote - fractal analysis and it worked out not only for me, here you can be absolutely sure of reliability.
it is not clear what and how you analyzed
My fractal analysis took me more than a year. Of course, I didn't do it every day, but it was tight enough, if possible. I'm not going to post everything, sorry.
and what's with the evasions.
I gave a definition of the term "evasive trajectory", maybe not a good one, but it's a standard term.
P.S. as for the phenomenon on the first page - double-checked and didn't get it. Have you tried checking on quotes from other DCs and/or other software?
Show me in detail what you did and what did you fail? Take M15, UERUSD for 10 years, maybe five. Take the Open(n)-Open(n-1) process, build histograms with different steps and show them here. But mark it with dots, not bars. It will be all clear, and so, not so much, I can say that the earth is flat and I will be right...in some sense.
I forgot to add to my post:
Thesame effect of highlighting the alpha and omega processes can be obtained by dividing the quoted incremental flow into, what is "less lambda" and what is "more lambda". Where lambda is an estimate of the boundary of the long tail of the incremental distributions. The exact lambda estimate is hard to find, but that's how I first got the two processes (for M15 EURUSD L would be about 2-3 RMS):
PS: Try it if anyone is interested (I just won't post it until Sunday at the earliest). Impressed by what you see :o)
That, imho, is a much nicer recipe for getting the phenomenon than the original one (using Matkadian histogram function as a black box). But it's not without dark spots either, so I guess we'll have to wait until Sunday.
I immediately remembered alsu and the quantile regression he promotes.
Farnsworth, no offence, I'm judging by my own experience, even though I don't own any such materiel. Maybe this will help. Just sensible stuff you traditionally write, but all the time it feels like half a step away from shifting the research to trade laths. The net balance in hindsight from the research is equal to or close to zero as a result. Often good researchers with a mathematical background become hostages to their knowledge themselves. You always want and can dig deeper, where you end up with so many assumptions and exceptions that the primary concept becomes unusable for trading. If it helps, try to look easier. I found a "pattern", I should try to trade it. It has become very unclear, so it should be discarded. The purpose of the money. And often the goal turns into the research and it is no longer a work with money, nerves, etc., but a hobby. That's not good. Sorry for the lyricism, maybe it will help.
To your observation, let me be honest, plain and simple. I only see the market as a business, which means I have a priori high requirements for the quality of the Product I create and of course, it's not my core business right now. Yes, there are successes, and good ones (which I put in real mode during testing), and I could shake off trimmed reports like paukas and write everywhere, that "it's like you can not" etc. But I'm honest about my achievements and understand that this is not yet a business (in my mind about it). Put up a tent on Tverskaya (for example), it will earn you money? - Very much so. The risks? Of course there are, but they are negligible compared to the almost random chain of events.
You see, I'm not a gambler by nature, that's why I act differently, I won't take any chances. Many many many years ago, when I owned the fine art and was a TA AND VA apologist - I believed and was cool, until I realised what edge of the abyss I was walking on, and even blindfolded (ahem, once again kicked TA/VA). Yes, the problem is complicated, but you write so that all 99.999% of forum participants have already developed such robust systems (pitched tents at Tverskaya), only I am left:o) Well ... what can I say, I'll try to catch up with my colleagues, to overtake them - and to take away the mall, er ... one is not enough - two malls and all the tents :o)