Machine learning in trading: theory, models, practice and algo-trading - page 1110

 
Grail:

the main question is HOW it happens

imho, not how but when it's more interesting)))

I wrote it here and not here and there were similar messages: the price moves within the day, closing the nearest historical maximums / minimums (High/Low), this creates the effect of a normal distribution, and then a bang-bang happens and the price, instead of repeating the movement it made during the nearest hour (or a different time interval), moves in one direction, thus destroying the previously created normal distribution

I am going to find a statistics program, it used to make a nice picture there

 
The Grail:

They generally float, in all dimensions, in linear time and in its scale, from instrument to instrument, etc. The main question is HOW it happens, what is the functionality of changes in statistics, in particular how regular are the change functions, if statistics are not changing continuously (at least piecewise constantly) and either regularly, then trouble, then only insiders are left to enjoy the market.

Speculation gets squeezed out.
If we are not interested in arbitrage, then only investing, "buy and hold", but it is a long time, but we want to eat every day, inside is not about us.
 

Such "curves" will be in all market prices, the instrument is not important, the TF is not important, the principles of processing are not important if you can reconstruct the price series inversely (I used logarithms and increments... it is not important)

There is an article in the forum "about the coin trend" somewhere, imho, there is no more adequate mathematical model of the markets :)

 

I found a method of instant optimization and instant prediction

Past prices are an arbitrary set of 1 to 9 digits and future prices are an arbitrary set of 1 to 9 digits.

The prices themselves, are formed, through mathematically impossible prices.

And yes, Pi equals 1.



 
mathematicgraal:

I found a method of instant optimization and instant prediction

Past prices are all possible sets of 1 to 9 digits and future prices are all possible sets of 1 to 9 digits.

The prices themselves, are formed, through mathematically impossible prices.

And yes, Pi equals 1.

Can you explain the philosophy of the idea in more detail?

 
mytarmailS:

Can you be more specific, tell the philosophy of the idea

he won't answer, he's already banned )))

I don't even want to look at the source code, judging by the name and the "curvilinear" picture it's another version of Fourier series decomposition into harmonics (you should set from which bar to build) and then the harmonics reconstruction on new data (new bars)

the original version has been lying in the kodobase for 6-7 years; the indicator itself is cunningly designed, it will even erase its "curves" in the strategy tester and constantly re-draw new ones

I made a version of this indicator a couple of months ago to order, so it wouldn't redraw old data in the tester, I even made an EA with one fixed lot... The percentage of prediction is really nil!

SZS: If I find it, I'll give the version for the tester without re-drawing, or tweak it yourself ...

 
Igor Makanu:

He won't answer, he's already banned )))

I don't even want to look at the source code, judging by the name and the "curvilinear" picture it's another version of Fourier series decomposition into harmonics (you should set from which bar to build) and then the harmonics reconstruction on new data (new bars)

the original version has been lying in the kodobase for 6-7 years; the indicator itself is cunningly designed, it will even erase its "curves" in the strategy tester and constantly re-draw new ones

I made a version of this indicator a couple of months ago to order, so that it wouldn't redraw old data in the tester, I even made an EA with one fixed lot... The percentage of prediction is really nil!

SZS: If I find it, I'll give the version for the tester without re-drawing, or tweak it yourself ...

Yes, that prediction doesn't look like a harmonic signal, maybe it's not that simple ...

 
Vizard_:

As long as the blacksmith doesn't turn out to be a... Econometrician.
Let's generate some curves and put them in a vr.

Let's try classical decomposition.

.........

That's cool, I never thought of that before.

 
Vizard_:

Fa, let's talk about the practical use of GARCH... Let's have a laugh...

I don't know when it will be, although I did the hardest part: choosing a model. I think it is quite an alternative direction compared to MO. GARCH is a mainstream Expert Advisor, unlike MO.


I haven't completed the testing at GARCH, because I've created a high-quality Expert Advisor in MO. But since June I haven't had enough time and effort to improve it to the commercial level.

 
Vizard_:

It is a natural desire to dig out and see what the cotier consists of. Before that, it is quite
It is logical to try to assemble it (SB topic) and check the kotier. Generate, unscrew,
to look from a different angle can be different. Below is a different decomposition...

To parse a quotient, you need all the trades and their volumes.
Reason: