What makes an unsteady graph unsteady or why oil is oil? - page 4
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
З.Ы. просто меня часто убивают высказывания это....вставить любое слово-неработает. Спрашиваеш почему. Ответ у меня куча лосов и нет прибыли ((( и человек не понимает. что такое утверждение бред. Если бы фурье неработал, то извините у Вас сейчас бы небыло ни копьюторов. ни телевизоров, ни сотовой связи....оглянитесь вокруг. Эти предметы есть ? значит он Фурье работает, только Вы его неправильно применяете...
I agree: "it doesn't work because it didn't work for me" is not an argument.
Your examples have something that is not on the market: the carrier frequency is periodic - which is why Fourier works for such systems. Similarly with noise suppression, for example in radar: it, the noise, is not periodic (for all systems). By isolating the carrier you can then isolate the useful signal.
IMHO, alas this is not for FOREX - too bad by the way ;)...
Good luck.
so it's Fourier that works, only you're applying it wrong...
If Fourier doesn't work, then all the maths doesn't work, although I could be wrong. The fact that I'm not applying it correctly, or not completely, is more likely.
"не работает" конечно зависит от человека. У меня - плохо. У других - получается хорошо. Буду изучать проблему.
Что вы можете посоветовать по поводу сигнал\шум? К фурье что-то нужно добавить, а что я не пойму. Фильтр, но какой?
Do a windowed Fourier transform of each bar and then trace the fate of each coefficient as a data buffer, then it will become clear what is meant by the word non-stationarity.
Completely stationary will be the data that will give the same coefficients all the time with this window shift
(i.e. by tracking a single coefficient you will see a straight horizontal line).
Prival is right that without Fourier there would not have been many breakthroughs.
Hence a reasonable question: what are the limits of the method for extrapolation?
It seems to me that you can predict a process with interpretation of Fourier transform,
but only for the process in which stationarity variations within prediction range can be neglected (and it depends on discreteness).
If I walk on the road, it is impossible to predict my movement to an external observer with a discreteness of one step,
as the next step can be in 4 different directions with the same probability,
but at a sampling rate of 25 frames per second it is elementary (but only for the inertia of my body).
IMHO the quotes have too huge discreteness to consider the process as conditionally stationary.
..., например, в радиолокации: он, шум, не является периодическим (для всех систем). Выделив несущую можно потом выделить полезный сигнал.
ИМХО, увы это не для FOREX - кстати, очень жаль ;)...
Удачи.
Glad to meet a fellow radar enthusiast. There is another method. Do not select the carrier, but immediately suppress, before processing, the output is Doppler frequency+noise. Determine the amount of noise. You may do it there (you are right, but there are problems in Forex).
Good luck too.
Ah, what exactly are you interested in?I'm interested in the graph from a mathematical point of view, if you like:). I.e. what it is in mathematical terms, understanding and concepts. Instead, there's a post-hardcore going on:) ...... You see, at the end of the day it all comes down to the chart, the whole macro-micro_whatever_ro economy, all the news noise and information flow, MarketMakers games, strikes, demonstrations, great victories, etc are all reflected on the chart and create it. Now, what do we know about this chart and are we paying attention to it? From the perspective of its fundamental understanding, i.e. what it originally is. OK, now I know that it is called an unsteady graph, expressing the same process. And even more: I understand the essence of this process (non-stationary)! And now with a bit of reasoning, certain things can be understood accurately and reasonably. For example: the special case approach to trading is nothing, but the general case approach is everything.
....
ИМХО у котировок слишком огромная дискретность чтоб рассматривать процесс как условно стационарный.
It's been a while since I've been on the forum. I see it's a pity to have someone to talk to. I also noticed it and even calculated (somewhere here on the forum) which fluctuations we can track. I tried to go with ticks, fiddled with them for a long time. I even ordered the tick assemblers. I was working on some of them myself, but once I understood one thing, my eyes just opened. The sample rate is not constant, and I used to think it was a constant (((. Fourier methods and spectral analysis in general imply that delta t is constant, but it's not, and everything started to fall apart like a house of cards.
I even made a thesis )) sampling rate rules the world )).
Tried to go into ticks, spent a long time fiddling with them. I even ordered tick assemblies. I was finishing some things there myself, but once I understood one thing, my eyes just opened. The sample rate is not constant, and I used to think it was a constant (((. Fourier methods and spectral analysis in general imply that delta t is constant, but it's not, and everything started to fall apart like a house of cards.
I even made a thesis )) sampling rate rules the world )).
You are not the first to encounter this problem. The solution is the right google queries and the right literature.
The simplest approach I've encountered is to collect ticks and their arrival times, perform linear interpolation between consecutive pairs of ticks, taking into account time. It is further obvious.
You are not the first to encounter this problem. The solution is the right google queries and the right literature.
The simplest approach I've encountered is to collect ticks and their times of arrival and perform the linear interpolation between successive pairs of ticks taking time into account. It's further obvious.
The right literature, can you elaborate on which one? It's not obvious to me personally, can you elaborate? You did a linear interpolation of the tics, then what?
I even put forward a thesis )) sampling rate rules the world )).
If you're not just talking about price changes? Are you suggesting that this also "works" in other "industries"? Of course, I'm not talking about radio engineering now, it's obvious there.
Простейший подход, с которым встречался - собираем тики и время их прихода, выполняем линейную интерполяцию между последовательными парами тиков с учетом времени. Дальнейшее очевидно.
The best variant is cubic spline approximation + ADC noise filtering (low bit error) directly at the input.How I hoped MT5 will store ticks. I hope it will store ticks in it. Renat did not write about it that it is important for processing, but he didn't listen. I am very sorry.