From theory to practice - page 1047

 
Uladzimir Izerski:

If you look at the logic, you can see your attitude towards your interlocutors.

You don't even hide it))

Everyone is such a sheep and I'm so clever!!!


You're projecting. Talk to a psychologist.

These are characters from the famous cartoon "Sean the lamb", in which the lamb is the smartest.

 
Maxim Dmitrievsky:

well, validate them, of course. If a pattern repeats often, why not an internet. Let's say weekly, monthly, daily returnees. I would rotate them in various combinations, obtaining some predictable periodic patterns. Recalculate occasionally.

Or is it going to be like a Fourier?

Fourier big calendar/periodic cycles are bound to look out. Once a year everyone tweaks (some tweak) the totals, corporations have regular surprise quarterly reports and the like. They will happen, and they are there...

It's supposed to be detected and subtracted...and work with the remainder...something like that I guess.

 
Aleksey Nikolayev:

You are projecting. Talk to a psychologist.

These are characters from the famous cartoon "Sean the lamb", in which the lamb is the smartest.

Did you choose the screensaver consciously or subconsciously?

Such seemingly simple things characterise a person.

 
Uladzimir Izerski:

Did you choose your screen saver consciously or subconsciously?

Such seemingly simple things characterise a person.

He has expressed his essence with these rams - he is stubborn and does not want to understand.
 
Maxim Kuznetsov:

Fourier big calendar/periodic cycles are bound to look out. Once a year everyone tweaks (some tweak) the totals, corporations regularly have surprise quarterly reports and so on. They will, and they are...

It should be detected and subtracted... and work with the remainder... that's probably it.

well here you can find more local cycles by subtracting some cumulative return from the price... it's not obvious, but what we do is we take out the linear trend and then we look for something in the residuals

by simple brute force.

 
Maxim Dmitrievsky:

I'd like to write some sort of enumerator, like you did in the last article about gaps... but something like that. Or I could just try arbitrary price patterns and draw some statistics.

Gaps are good because of their precise timing. And take profit is as accurate as a drugstore. Any other patterns are much more vague, which forces us to switch to random statistics, which is a difficult and poorly developed area.

 
Maxim Dmitrievsky:

well, validate them, of course. If a pattern repeats often, why not an internet. Let's say weekly, monthly, daily returnees. I would rotate them in various combinations, obtaining some predictable periodic patterns. Recalculate sometimes.

Or would it still be Fourier? Or not. And if you introduce the ratio of price to cumulative return with a chosen range of lags and that ratio find stationarity, peace and tranquillity

For some reason, I don't believe in Fourier (and its inherent wavelets and fractals). Technically speaking, they can be applied to non-stationarity, but that's not the stationarity we need. Roughly speaking, by multiplying a stationary series by a cosine, we get a non-stationary series, which may well be decomposed in Fourier, but it has nothing to do with prices - rather the voltage in the socket)

 
Aleksey Nikolayev:

For some reason I don't believe in Fourier (and its inherent wavelets and fractals). Technically speaking, they can be applied to non-stationarity, but this is not the stationarity we need. Roughly speaking, by multiplying a stationary series by a cosine we get a non-stationary series, which may well be decomposed in Fourier, but it has nothing to do with prices - rather with the voltage in the socket)

So it's time to call it a day ))

 
Aleksey Nikolayev:

For some reason I don't believe in Fourier (and its inherent wavelets and fractals). Technically speaking, they can be applied to non-stationarity, but this is not the stationarity we need. Roughly speaking, by multiplying a stationary series by a cosine we get a non-stationary series, which may well be decomposed in Fourier, but it has nothing to do with prices - rather with the voltage in the socket)

The bogeyman of non-stationarity haunted him.

Actually stationary series - once, and that's it.) Or a little more than that. And nothing, everyone is alive and doing fine with it. Same radio amateurs, by the way.))

 

The branch is a cart, and the forum members are a goose and a pike)

Reason: