Pair trading and multicurrency arbitrage. The showdown. - page 121

 

where are you trying to get trading volumes of currencies from?

they don't exist... nobody publishes them, they can be calculated with all possible data only after-the-fact, with a delay of a shitload of time.

immediately: and do not cite CME public, because 1) currencies are not their speciality, they are American fund 2) address transactions are much bigger and exchange volume (which you confuse with turnover by the way) is nothing against their background.

 
Roman Poshtar #:

Good link but it doesn't work on one dts. What if we make a program to track 10 dts for example? What do you think? Will they sleep?

What exactly do you want to track? There are 2 problems there:

- NO sufficient spread ideally (when full balance, in the example by volume means 1-1-0.8658);

- If we take a real example, we need to open a third position with either 0.86 or 0.87 volume. This creates an "artificial bifurcation", and it, in turn, may not converge;

The details are on the screen. Above is the ideal. Volume is balanced on each bar (equal to EURGBP price). On the bottom is the case where we took a volume of 0.87 for GBPUSD. The spread will converge at the moment when the EURGBP price is 0.87. And this may not be the case.


 
Roman #:

Yes, it's clear, you can do it that way, it's more with the code just strapping.
It's not the main task for now.

Roman, I understand.

Only those who can think outside the box will understand everything that is described.

Wasted years put a certain stamp on a person's logic.

This is bad, as it hinders intellectual development.

Almost no one wants to create and search.

Only find your own, bring it and give it in your hands ;)

 
Renat Akhtyamov #:

Roman, I get it.

Only those who are able to think outside the box will understand everything that is stated.

Wasted years put a certain stamp on a person's logic.

This is bad because it hinders intellectual development.

Almost no one wants to create and search any more.

Only find your own, bring it and put it in your hands ;)

I am more surprised that venerable guys can't grasp the meaning, for whom data preparation is the first postulate.
The fact that you build different curves, I understood it at once. I just built an intraday, so to speak, to work out a template from the code and make a correct calculation.
When the correct calculation is made, you can modify it to suit your needs. But so far I can't work with the formula, I have a rough idea of what's wrong,
, but I'll show you everything later in a private message.

 
Roman Poshtar #:

Your topic has grown a lot.

Please tell me, as of today, have you managed to create a working algorithm of pair trading without resorting to averaging, martin or over-sitting?

 
Ivan Butko #:

Your topic has grown a lot.

Please tell me, as of today, have you managed to create a working algorithm of pair trading without resorting to averaging, marting or over-sitting?

No, it has to have a deposit of $1000
 
Roman #:

You have an understanding of data retraining, don't you?
It's the same thing. Only the output will be not just something, but stationary true data without errors.
You can already try to trade this, and those who are stronger can go further in development.
It is strange that you are dumb in this case.

One more Katshchikophile, then it is not even worth talking about, it smells like logic from a mile away.

 
Maxim Dmitrievsky #:

One more Katushchikophile, then it's not even worth talking about, you can smell logic from a mile away.

Ugh on you, stinking dog...
Your favourite Prado rests from the simple formula of Rena ))

 
Maxim Dmitrievsky #:

already reeks of logic from a mile away.

It's brilliant)))))))))))
 
Roman #:

I'm more surprised that venerable guys can't grasp the meaning, for whom data preparation is the first postulate.
I understood that you build different curves right away. I just built an intraday, so to speak, to work out a template from the code and make a correct calculation.
When the correct calculation is made, you can modify it to suit your tasks. But so far I can't work with the formula, I have a rough idea of what's wrong,
, but I'll show you everything later in a private message.

Roman, have you designed digital filters on microchips?

If not, I'll tell you the following.

MO is a 100% digital filter, where the predictors are the same coefficients of the digital filter.

What logic can there be in filtering money where every fraction of the price matters?

The guys have gone so far into the wilderness that they can't crawl out now, they are lost ;)