From theory to practice - page 280

 
Yuriy Asaulenko:

I forgot, do you work on Forts? So there is no impact (practically none) if you don't go beyond reasonable volumes.

If you don't go beyond reasonable volumes, it's almost invisible.

Well, I'll see it anyway.

even 0.01 on cents rules the price, or appears for a reason

 
Renat Akhtyamov:

If you don't go out, it's almost invisible.

Well, I'll see it anyway.

For this reason I do not know whether I will start a new one or not.

The person who knows about my inventions has already moved on to another forum.

I don't know much about the DCs. The quotes there are not market ones. I do not see how our trades can influence anything there. On the other hand, DT has a full right to quote instruments. Figure it out)).

However my brokerage company closed for RF citizens on 23.03. I will start a new one or not - have not decided yet.

 

Read the first post of this thread for the first time.

I 100% agree, the process is non-Markovian and I noticed it a long time ago, but didn't know where to start.

Thank you, moving on...

With the aim of automating this approach, I'll try to move this way:

Files:
Fund_22.zip  175 kb
 
Renat Akhtyamov:

Read the first post of this thread for the first time.

I 100% agree, the process is non-Markovian and I noticed it a long time ago, but didn't know where to start.

Thank you, moving on...

I will try to go that way:

The process, however, is Markovian. I've already written that there is nothing random in the market - all actions of traders are subject to the algorithm if...then... No normal trader does anything randomly. Another thing is that there are a lot of traders.

It also applies to forex, because forex quotes are generated by the real market.

 
Yuriy Asaulenko:

The process, however, is Markovian. I have already written that there is nothing random in the market - all traders' actions are subject to the algorithm if...then ... No normal trader does anything randomly. Another thing is that there are a lot of traders.

It also applies to forex, because forex quotes are generated by the real market.

Who's arguing? Here you are:

A non-Markovian process is a random process whose evolution after any given time t {\displaystyle t} depends on the evolution preceding that time. In other words,the "future" of a non-Markovian process depends on its "past". A non-Markovian process is a random process with memory, whereby by speaking of the memory of the process, it is meant that the nature of the evolution of the process in the past determines its statistical characteristics in the future. A non-Markovian process is opposed to a Markovian process.


It's not clear when and to whom it will occur to open buy or sell. It's a random process. But the response of forex (the adaptive system) will be in some time to this order anyway.

 
Renat Akhtyamov:

Who's arguing? Here we go:

It's not clear when or who will think of opening a buy or sell position. It is a random process... But the market response after some time to this order will still happen.

Some pages ago there was a post of mine that quoted Heisenberg.

It doesn't matter that a lot of people make decisions independently. Nevertheless the process will remain Markovian - with memory etc. Well, or as-is-Markovian).

If you and I somehow manage to get under the skin of all market participants, we will inevitably reproduce all quotes almost exactly. starting from the time To, at least on some time interval.

The question is different - what is state? State is a vector - (x1,x2.....,xn). If the state is set correctly, then we identify the process correctly. Suppose we take as a state a vector -(x1,x2.x3) and don't take into account other parametres. The second vector would appear to be unmarked and without memory and have nothing to do with the process at all.

In addition, I can say that even with simple methods you can successfully predict the market for 70% of time interval. If a process had no memory and its state were independent of the previous ones, it would be impossible to implement it.

 
Yuriy Asaulenko:

There was a post of mine a few pages ago quoting Heisenberg.

Never mind the mass of people making decisions independently. Nevertheless the process will remain Markovian - with memory etc. Well, or as-is-Markovian).

If you and I somehow manage to get under the skin of all market participants, we will inevitably reproduce all quotations almost exactly. starting from the time To, at least on some time interval.

The question is different - what is state? State is a vector - (x1,x2.....,xn). If the state is set correctly, then we identify the process correctly. Suppose we take as a state a vector -(x1,x2.x3) and don't take into account other parametres. The second vector will appear as unmarked and without memory and it won't have anything to do with the process at all.

In addition, I can say that even with simple methods you can successfully predict the market for 70% of time interval. If a process had no memory and its state were independent of the previous ones, it would be impossible to implement.

I think that it's wrong to draw conclusions about the Markovian character of the process based on the actions of market participants. But the fact that the market has a memory is certain. For example, the market remembers levels well enough and for a long time. I'm not sure that it's true and correct, but I would say that 80% of price movement is non-Markovian and the remaining 20% are Markovian. I don't think price behaviour is entirely determined by previous history. Otherwise creating a grail (predicting with 100% probability) would be a solvable task.

 
Yuriy Asaulenko:

There was a post of mine a few pages ago quoting Heisenberg.

It doesn't matter that a mass of people make decisions independently. Nevertheless the process will remain Markovian - with memory etc. Well, or as-is-Markovian).

If you and I somehow manage to get under the skin of all market participants, we will inevitably reproduce all quotations almost exactly. starting from the time To, at least on some time interval.

The question is different - what is state? A state is a vector - (x1,x2.....,xn). If the state is set correctly, then we identify the process correctly. Suppose we take as a state a vector -(x1,x2.x3) and don't take into account other parametres. The second vector would appear to be unmarked and without memory and have nothing to do with the process at all.

In addition, I can say that even with simple methods you can successfully predict the market for 70% of time interval. If a process had no memory and its state were independent of the previous ones, it would be impossible to implement.

That's what I'm going to check now.

I've read a lot of literature...

Checked by the mother-mate - Markovian/non-Markovian

 
Renat Akhtyamov:

That's what I'm going to check out now.

I've read too much literature...

Checked by the way - Markovian/non-Markovian

Rena, since you're a reader, please read something about non-entropy. IMHO this is the parameter responsible for the "trend/flat". References to literature (no matter - in English or in Russian) on this topic - please!

 
Alexander_K2:

Rena, since you are a reader, please read something about non-gentropy. IMHO this is the parameter responsible for "trend/flat". Literature references on this topic - please!

You gave me the links, they're above.

And by the way, I haven't read as much on your topic as you and apparently I'm just starting out.

Now I've been digging into the Wiener process. If Wiener, then the cotier by non-Markov, like something like that, but I'm not sure yet.

Although, here's one for Alexey, but it says it's Markovian again with examples for the exchange

http://pandia.ru/text/77/396/100591.php

I'll figure it out now... I don't understand anything yet.

Лекция 13. Винеровские процессы и лемма Ито
Лекция 13. Винеровские процессы и лемма Ито
  • ОриентМикс
  • pandia.ru
Лекция 13. Винеровские процессы и лемма Ито 13.1. Марковское свойство ……………………………………………………………. 2 13.2. Стохастические процессы с непрерывным временем ………………………. 3 13.3. Процесс, описывающий изменение цены акции ……………………………... 8 13.4. Параметры ……………………………………………………………………… 13 13.5. Лемма Ито …………………………………………………………………………13 13.6. Свойство...