Filter without delay - page 13
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
Also understand that we can judge in different "environments" you judge "for one" I judge "for another". For example you judge for daily timeframes and I judge for weekly timeframes. :) So we should clearly define the "environment" ("environment").
I'm talking about the approach. I must admit that I'm not ready to do it myself. No need to talk about Fourier at all - it will go on and on - the prediction is obvious. What I have seen from the wailets. They are hillocks, one of the coordinates of which is time. The prediction is that we have not moved out of the knoll and the found frequency of the wavelet will continue to exist for some time. Changing of this hill along the time axis will provide the ground for entering or exiting the position. Since the task cannot be set in the stationary markets at all.
Frankly, I was also interested in this question. But it turns out that there is a filter (a set of filters that transforms the market into a straight, growing line. Is such a formulation of the question realistic? By doing so we deny such a property of the market as "uncertainty". Although there is futures on Bernanke, but what to do with Negroes?, earthquakes, and more likely simple market manipulation and petty DC cheating? Not a realistic ideal.
All the things we don't (can't) account for create variance in the results. A more realistic model of an ideal TS))) is SB with positive drift. The ratio of this drift (mo) and CS characterizes the quality of the system. This is how the Sharpe coefficient is calculated, for example. But this is all temporary. Sooner or later, any trading idea and the system based on it will stop matching the market, and its equity will become non-stationary, like BP itself. And no matter how hard you try, there are no eternal methods. There is no such thing as a doughnut that can be adapted to any market at any time. And it's not worth wasting time on. There are time specifics.
The very first consideration with which to start building an adaptive filter: there are two parametres "smoothness (number of kinks)" of the filter Error1 and the deviation of the filter from the price Error2.
For example, if Filtr[i]=Close[i], then Error2=0, Error1=x; if Filtr[i]=1 for all i, then Error1=0, Error2=y. The truth is somewhere in the middle, and it changes with time Error1=f(Error2, time) and so we need to study the function f through other price invariants hence the initial question: why do we need filters at all, is it easier to study other "price invariants" at once?
And what is a signal in the market? a group of bulls?, a Trend? Even if you identify a "signal", there is no guarantee that it will end and then what is there to adapt to? There are plenty of adaptive indicators. Big masters but calls them "toys" for some reason.
All that we do not (cannot) account for creates variance in the results. A more realistic model of an ideal TS))) is an SB with positive drift. The ratio of this drift (mo) to CS characterises the quality of the system. This is how the Sharpe coefficient is calculated, for example. But this is all temporary. Sooner or later, any trading idea and the system based on it will stop matching the market, and its equity will become non-stationary, like BP itself. And no matter how hard you try, there are no eternal methods. There is no such thing as a doughnut that can be adapted to any market at any time. And it's not worth wasting time on. There are time specifics.
No MO and no variance - why discuss something that doesn't exist.
No MO and no variance - why discuss something that doesn't exist.
The results of the system should be more or less stable. What is the point of trading a system if there is no certainty that it will make a profit? Better not to trade at all then.
the results of the system should be more or less stable. What's the point of trading a system if there is no certainty that it is profitable? Better not to trade at all.
OK
Вот что фикция - так это теории вокруг японских свечей. Надо бы написать прогу по проверке всех этих фигур, прогнать на истории, собрать статистику успешных/неуспешных предсказаний и разоблачить японских шарлатанов ))
:-)) Quacks, indeed!...
The Japanese invented this system before there was forex. And there were no speculative markets on the planet. That was centuries ago.
And it was all about predicting the rice crop. One candle = a year. Now we just use their workings on all market curves.
Шут с ней, с крутизной. Я согласен быть самым некрутым на форуме.
Почти год я пытаюсь доказать бесперспективность любых ТС, связанный со словом ЦОС, нормальный закон, Фурье и т.д. Причина банальна: этого нет на рынкете. Сам рынкет - это другой объект, не сводимый к стационарному процессу. У него конечно имеются стационарные участки, но ИМХО нада заниматься тем объектом, который имеем, а не менять объект к своим знаниям.
Это относится практически ко всем участникам форума. Когда-то самый крупный местный авторитет написал "вейлет - это еще тот лохотрон". Как всегда без аргументации, но на то он и авторитет, что хорошо знает всех бегемотов на форуме.
Ринкет - не стационарный процесс. Более того: неопределенный процесс, т.е. события, которые влияют на котиры, которые нельзя предстказать (война в Ираке).
Большинство ТС долго не живут и причина одна - поменялся рынок, поменялась периодичность рынка (поменялся период машки и любого другого индикатора). Никто не умеет ловить момент этого изменения. Моожно привести к прибыльности ТС за счет оптимизации (заклеймлено полностью), но нет гарантий, что пока оптимизируешь, рынок снова не поменяется.
Это похоже на частотную модуляцию, но там имеется закон, который можно распознать. На рынкете частота меняется, но я не слышал чтобы кто-то сумел подстроиться под это.
Бояться самого матлаба не следует. Часть работы сделал Рашид (он не знает об этом) - описал выход из MQL5 в DLL на С++. Сам матлаб имеет С++ и строит DLL библиотеки. Имея этот переход становятся доступными все функции матлаба. А там можно будет взять если не уменьем, то числом. По вейлетам существует огромная отечественная литература, но она в геофизике и кардиологии
I would like to disprove this...
The market can be reduced to a quasi-stationary form using PF or similar decompositions.
Here, what's not stationary?
I would like to disprove this...
The market can be reduced to a quasi-stationary form using PF or similar decompositions.
Here, isn't it stationary?
Have a look at the attachment PDF file in Russian at your leisure. I will not teach anybody else, read and discuss.
Посмотрите на досуге аттач PDF файл на русском. Учить больше никого не буду. читайте и будем обсуждать.
I can't see how a wavelet transform can help in forecasting...? Or just in trading?
I have almost harmonic curves in the pictures. You can already use them to make predictions. What's the wavelet for?