Machine learning in trading: theory, models, practice and algo-trading - page 1098

 
Maxim Dmitrievsky:

So if the market is random too. )

I just need to understand why.
 
Igor Makanu:


1)-we can't guess when a major player will show up? - surprise?

2)-we can't guess the big player's target? - Today 5 pips, tomorrow 10 pips - unexpected?

3) - we look at the charts of currencies thinking that the players' goal is to make a profit on the price movement, but players often come to the interbank market to buy currency and leave the market, the goal is just an exchange, not price movement - unexpected?

4) Well, about the TF, let's go back to examples: here's an internal combustion engine, we do not know its principles, and the sequence of ignition of the mixture 2-3-4-1- 2-3-4-1 does not seem logical and ... random? OK, we take TF M5 for this sequence, we got 1-1-2-2-3-3-4-4

well, what do we understand what an internal combustion engine is? - no, it was an incomprehensible mechanism for us, and still is, but we are sure that now it is not random.... and then engine frequency changed and instead of logical 1-1-2-2-3-3-4-4 we got 1-4-2-3-1-1-3-4 - again we need to pick up TF?

;)

1) when you appear, he will appear, he is your counter agent, and you are his

2) What is there to guess, it's your stop).

3) buy from whom? the air? see point 1)

4) I do not know about the time frame, I do not have a solution yet.

 
Novaja:
It remains to be seen why.

In terms of what?

It's just a lot of random factors

 
mytarmailS:

1) when you show up, he will show up, he is your counter agent and you are his

2) What is there to guess, it's your stop)

3) buy from whom? the air? see point 1)

4) I don't know about the time frame, I don't have a solution yet.

When you show up, no one will even notice, if you're not Soros

 
MytarmailS:

3) buy from whom? the air? see point 1)

I mean the number of participants and their goals we do not know, who entered the market and waits for profits on the price difference, and who entered and exited the market having bought at the current price, while the first participant is a speculator so to speak, he creates an additional market fluctuation when exiting the market - but it will be spread in time


Maxim Dmitrievsky:

when you appear, no one will even notice, if you are not Soros.

It's clear, but the mechanics of the market should be studied, the only thing we are given is the history of deals made, if there are repetitions, it's the market reaction and the actions of market participants... but from the history...

 
Igor Makanu:

It's clear, but the mechanics of the market should still try to study, all we are given is only the history of already made deals, if there are repetitions, then it's the reaction of the market and the actions of market participants... but in history...

the mechanics of the market is an auction, with inefficient ones you can see the patterns at a glance

on effective ones like handicaps you have to get out of hand and be the best of the best...

everything in auction history is a random process of self-organization of participants. Any market strives to become efficient.
 
Maxim Dmitrievsky:

the mechanics of the market is an auction, inefficient patterns are immediately visible

There's finally a ray of light in the kingdom!

I've also written about it in different words, that's why we see that prices update highs and lows, although a few days ago you said we should use price increments - what kind of auction is it? We'll consider an inertial model with speeds and momentum

In principle we probably need NS committees to evaluate two models - an auction model and an inertial model - they both work in the market, but which one is more efficient and when?

 
Maxim Dmitrievsky:

When you show up, no one will even notice, unless you're a Soros

Igor Makanu:

All right, but my point is that both the number of participants and their goals we do not know,

We don't know, I agree, but indirectly we can try to identify ...

1) the big player is a counter-agent of the crowd, he won't go in without the crowd, there is nobody to open and close a position with

2) The crowd works on the statistics "market memory" according to the principle - do it now, it was profitable to do it in the past

This is the moment when we can play with indirectly identify areas where the crowd will do something unambiguous.


For example here, the market has been saying to buy for almost the whole day, and we could already know yesterday that there will be a fall today

The green dots are overbought, the crowd was buying the whole day yesterday


It's very rare for the whole day

I have already showed the part of this indicator in work to the Wizard, but he did not take it seriously. I asked Maksim to help with MT4, but he too has other things on his mind.

I think it is possible to make predictions, but it is difficult and not always.

 
Maxim Dmitrievsky:

I am not good at this, but my theoretical background is weak in NS and lacks theoretical training.

And it's not about predicting price on increments (though you can), but about input/output classification on this data, in multidimensional space. I.e. approximation of points of a plausible regularity, stable on history for some time. This is not a deductive approach, i.e. we cannot say in advance what regularity is found... we are looking for purely mathematical plausibility, with the addition of some a priori assertions. And it doesn't matter at all whether it is a "real" law of nature or just a coincidence, since tests on new data put everything in place

I'm not good here, I'm not very good at NS yet, I have enough theoretical knowledge, I can teach NS and use it in the tester, but my theoretical knowledge is weak, one good man recommended me a fundamental reading of 1100 pages, I'll be reading it for a month yet

mytarmailS:

For example here, the market has been telling everybody to buy for almost the whole day, and we could know today that there will be a fall as early as yesterday

I wrote that we do not know the number and purpose of market participants, maybe there was no demand, or maybe there were buyers and sellers in equal proportions, or maybe we do not need to know it, the profit will not increase)))

mytarmailS:

I asked Maxim to help to transfer it to MT4 but he's busy with his own business too.

What is your source code written in?

 
Igor Makanu:

1) alas, until you check it all on the history is not a fact, I wrote that we do not know the number and purpose of market participants, perhaps there was simply no demand, and perhaps there were buyers and sellers in equal proportions, but it is possible that we do not need to know, these speculation will not profit more )))

2) What is your source code written on?

1) the indicator has been working for almost a year and I'm not the only one who looks at it

2) it is written on "R" I tried to rewrite it on mt4, it makes no sense to overwrite it on mt4, it's better to set up data exchange with mt4