What makes an unsteady graph unsteady or why oil is oil? - page 26
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
It's been a while since I've been here. Stumbled across this thread. It's an interesting discussion.
First question to the participants: why is the first price difference stationary? Has anyone calculated the moments of such a process?
The second and more important question: why do some people think that the stationary process is predictable? White noise is stationary too, but unpredictable. For those who do not believe, I can scientifically prove it. Or you can also do it this way. Imagine that white noise was predictable. Then the noise in the receivers wouldn't be a problem. Before receiving a signal we calibrate the receiver for external and internal noise and then at the moment of receiving a signal we begin to subtract the extrapolated noise from the noisy signal and get a clear signal. Shall we write a patent application together? :-)
I have written this indicator for calculating ACF for first price differences. I am attaching the code. I ran it on hourly EURUSD prices. The result is depressing: the correlation between the current difference and the previous differences is less than 0.01.
Первый вопрос участникам: почему первая разница цен стационарна? Кто нибудь рассчитывал моменты такого процесса?
The first price difference is not stationary. However, we can consider it stationary within the framework of some simplified model. Many propose to consider the first price difference as normally distributed, which is not entirely accurate, but may be acceptable again within the framework of the adopted model. Yes, thick tails. There are papers proposing to model price increments as a t-distribution with degrees of freedom of 4 or 5. It looks similar. For special connoisseurs, stable distributions, but already impractical. In any case, the first and third moments are zero, the second and fourth something...
Second and more important question: why do some here think that a stationary process is predictable? White noise is also stationary, but unpredictable. For those who don't believe it, I can scientifically prove it.
The question is what you mean by "predictable". In the context of the forum, I assume we want to make a profit. If you show me a tradable process that is white noise, I will become rich very quickly. The trading strategy, I think, is obvious - trade from the edges to the centre.
Написал вот такой индикатор расчёта АКФ для первых разниц цен. Прилагаю код. Прогнал его по часовым ценам EURUSD. Результат удручающий: корреляция между текущей разницей и предыдущими разницами меньше 0.01.
Greetings! Good to see.
It's been a long time, some of the stationarity tests are going on
I don't know who, it's my first time here in a long time, but it's absolutely true - stationarity doesn't make the process predictable.
I've already figured out how to make money from a random process. :о) If I set parameters of the forecast to make balance become a stationary process (it will be a random process anyway) then I may earn some money (knowing parameters of initial process). If you save money for an extreme drawdown and as soon as you get the largest possible profit (RMS of the balance can be determined) you may leave the market and never appear in the market.)
Nice to meet you here too. I am still at work trying to create a 'thinking network' close to the brain using splices and time. But so far, apart from simple object recognition, nothing works. Many people doubt that spike networks have any advantages over conventional neural networks. But that is a topic for a separate discussion.
If you show me the trading process, which is white noise, I will become rich very quickly. The trading strategy, I think, is obvious - trade from the edges to the centre.
Первая разница случайного блуждания это белый шум. Если можете получить прибыль на белом шуме, то с таким же успехом можете получить такую же прибыль на случайно блуждающей цене, ЧТД
По поводу частоты дискретизации наблюдаемого вр.ряда(тайм-фрейм), то
выбор тайм-фрейма(временного окна) оч. сильно влияет на спектр временного ряда,
но выбор этого самого тайм-фрейма это вопрос ...вкуса! :))) Шаманста или искусства, если хотите! :)
Потому что проблема выбора тайм-фрейма не формализуема и опр. личными предпочтениями трейдера.
Но уменьшение масштаба, частотного диапазона усложняет статистич. картину, усложняет за счет
появления большего количества деталей.
The choice of timeframe should not affect the spectrum. You simply cannot analyse the spectrum above a certain frequency.
Below this frequency the spectra are the same. Do you need these HF details? If the energy of what is being discarded is quite significant?
Accordingly, if we extrapolate the BP of a random walk over a large number of n samples, e.g. using a conventional linear regression, then with high probability the extrapolation result beyond n samples will be close to the straight line n * (p - q)
at p=q, which will model the zero-mo increments, this line will be the abscissa axis (y=0)
of course, if p<>q the situation is different - the line will have a non-zero angle to the abscissa axis
the figure represents just such a case (p<>q) - "the asymmetric walking".
the bounding lines n(p-q-e) and n(p-q+e) - due to the increase of dispersion with the increase of the number of samples (n)
The choice of timeframe should not affect the spectrum. You simply cannot analyse the spectrum above a certain frequency.
On page 16 I gave spectra for M1 and H1 for the same time interval - 480 hours. The spectra are fundamentally different. The type of spectrum corresponds well with the physics of the process. Many trends on M1 started and ended in 1 hour, which corresponds to investors with different time horizon. This is probably the main reason for the non-stationarity of BP. Your statement should be perfectly valid for stationary Fourier decomposed series.
---------
Below this frequency the spectra are the same. Do you need these HF details? If the energy of what is being discarded is quite significant?
There are methods for averaging the SPM. That said, if the existing peaks differ little from the average, it's a flat. In trends, the main power should be concentrated in the peaks.
The first price difference is not stationary. However, we can consider it stationary within the framework of some simplified model.
In the Box model, one of the parameters is precisely the order of the difference. This value is determined at model identification and varies from 0 to n. Box states that for most models this parameter is no more than two. But restrictions are imposed on the difference coefficients. From Box, the model coefficient related to taking differences is determined by experience rather than science and it is not necessary to achieve stationarity of the resulting difference.