From theory to practice - page 1428
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Burnakov's latest posts on Habra are also on reinforcement learning, which means that the models are quite clear.
That is, it is pure AI that is trying to understand the market itself.
But it's not clear if it's having any success. Personally, I have so-so in this direction so far.
My model is also known to everyone except Asaulenka and Makanu. The market is a random process with memory.
And if everything is clear with random process - for its description there is direct Kolmogorov's equation and simple probability theory, then it's more complicated with memory...
There are wonders in the market - these are non-linear time intervals between quotes, the so-called Erlang distribution, which uniquely indicates the presence of consequence in the tick flow and an unusual form of probability density of increments, etc., etc., which add difficulties in studying....
Only the figurative Asaulenki are frightened by this and just get cold feet.
One can, of course, spit on memory and treat the market as a random process. The conceptual question is whether it can be beaten? The answer is yes. I don't know if a neural network can do it, but probability theory and DPT can handle it.
The model I have is also known to everyone except Asaulenka and Makanu. The market is a random process with memory.
And if everything is clear with a random process - there is a direct Kolmogorov equation to describe it, and simply probability theory, then it is more complicated with memory...
There are wonders in the market - these are non-linear time intervals between quotes, the so-called Erlang's distribution, which unambiguously indicates the presence of consequence in the tick flow and an unusual form of probability density of increments, etc., etc., which add difficulties in studying....
Only the figurative Asaulenki are frightened by this and just get cold feet.
One can, of course, spit on memory and treat the market as a random process. The conceptual question is, can it be beaten? The answer is yes. I don't know if a neural network can do it, but probability theory and DPT can handle it.
MO is also a terver in essence, of course it can, but you have to know how.
For example, I successfully use MO for arbitrage, but for raw data I somehow cannot completely master it. This is if you look at the tester. If in real life, of course, there may be good profitable months, but not always
non-linear time is the most important characteristic of BP
Non-linear time is the most important characteristic of fin.Once again, I would like to remind you of the words of Koldun (even though you sometimes get into violent altercations):
"The most important thing in MO is the preparation (preprocessing) of the input data. How this is done is the secret of every master of MO. My series is a combination of several currency pairs and time".
I don't know what multiple currency pairs have to do with it. I don't know... Maybe he read the Bablacocos thread? I have no idea...
But time, yes. How could he not? How does the net know that a given set of quotes came in at night and the same size in the day? Without time, we just have a stupid set of random numbers - SB. And it is extremely difficult to deal with. But, you can.
Once again, I would like to remind you of the words of Koldun (even though you sometimes get into violent altercations):
"The most important thing in MO is the preparation (preprocessing) of the input data. How this is done is the secret of every master of MO. My series is a combination of several currency pairs and time".
I don't know what multiple currency pairs have to do with it. I don't know... Maybe he read the Bablacocos thread? I have no idea...
But time, yes. How could he not? How does the net know that a given set of quotes came in at night and the same size in the day? Without time, we just have a stupid set of random numbers - SB. And it is extremely difficult to deal with. But, you can.
He wrote that he doesn't have a stable TS. Anyway, sometimes it works, sometimes it doesn't... like everything else in life.
Are we discussing martin or me here? In all my time in forex I have not sold a single software product and have no intention of doing so. Ow! Is there anyone on this forum who can refute this?
Everyone has their own approach when preparing an Expert Advisor for trading. Personally I don't use optimizer because my EAs work on ticks and one test run from 2017 lasts about a day. So using the optimizer in the tester is not realistic, I will not live to see it through, I am 76 years old already). Constructing algorithm and defining parameters in the process of visual testing. There is no fitting.
There are over 500 trades here. Not enough? Maximum drawdown less than 7.5%. Loss is closed at 10% drawdown. In each such case I find out the reason and either correct the algorithm or adjust the parameters. So I'm not going to wait until some Expert Advisor of mine fails.
76 is a respectable age - I want to thank you for living so long and being sane!
After all, trading is an art, a game with people (their tools), there's nothing to state and even less to state axiomatically - like a pitchfork in water. If you are able to work with martin - great! I'm just stating my opinion as a result of experience and statistical experiments, but I clearly understand that even 1% of the possible is not declared and no one can embrace, so we are like prospectors looking for gold, digging ...
But, time, yes. How can you do without it? How does the network know that a given set of quotes came in at night and the same size in the day? Without time, we just have a stupid set of random numbers - SB. And it is extremely difficult to deal with. But, you can.
http://chaos.sgu.ru/kafedra/edu_work/textbook/khovanovs-01/node14.html
I understand everyone hangs out here. Guys, suggest a news indicator for MT5.
It takes years of programming and analytical work to create a quality news indicator.
The cost of such an indicator would be seven zeros.
You cannot afford such a luxury.
You can find a dud at a scrap yard, but it will be of no use to you.
The model I have is also known to everyone except Asaulenka and Makanu. The market is a random process with memory.
It is a random process with no memory.
It's enough to test a bunch of other EAs and see what is often referred to as a fit with history.
It's not a fit, it's - "The market is a random process WITHOUT memory".
It takes years of programming and analytical work to create a quality news indicator.
The cost of such an indicator would be seven zeros.
You cannot afford such a luxury.
You can find a hand-made one at a dumpsite, but it will be of no use.
Come on.
There is an economic calendar for this task and it is already built into MT5
just search the code base or order it from a freelancer