Machine learning in trading: theory, models, practice and algo-trading - page 1025

 
mytarmailS:

Well look, I told you that I will not give ready-made solutions, at least now .....

If in general, what do we all have in common in trade...? think about it...

In fact the only thing we have in common (besides fear and greed :) )

And this one thing is called statistics. The beginner trader looks through the charts, looking at what happened yesterday, the day before yesterday, a month ago, and the tough trader who is already programming, optimizing algorithms on the history, the first and the second traders are simply searching for some statistically significant things, as a result the market will "bump" both the first and the second, which is really funny, this will be a true pattern))

So with the help of neural networks or other MO you can simulate the actions of thousands of traders with "sufficient" accuracy, as well as the accuracy depends on the quality of the algorithm you wrote, i.e. predictors and target

What MO?

 
Alexei Tarabanov:

What kind of MO?

MoD algorithms - machine learning

 
mytarmailS:

MOI algorithms - machine learning

And the target one?

 
mytarmailS:

ME algorithms - machine learning

see what the deal is....

MO is the work of an algorithm

You may stumble on the fact that the output of the algorithm is predictable.

Such in the financial market gradually leads to a loss commensurate with the size of the deposit.

//The price up/down by a pips, a requote, the necessary market order is opened, then a pips to the downside, and there is another sponsor in the market...

MO is out of the question, for sure.

and I even thought you were working differently.

but the way is right

 
Renat Akhtyamov:

can stumble on the fact that the output of the algorithm is predictable.

Such in the financial market gradually leads to a loss commensurate with the size of the deposit.

well, why not use it?

Alexei Tarabanov:

and the target?

it is

 
What is a target?
 
mytarmailS:

I don't know)) I didn't watch the market on Friday at all, and it's important. My levels are drawn real-time, i.e. one hour or two or three hours before the rebound.

I have two types of levels:

1) just a usual local rebound, i.e. a slight movement within the noise

2) A trend level, that is a level that starts a trend, immediately or the next day, and only after a while this level will show itself on the chart

For example on Thursday the EUR trend started, I knew it would start on Wednesday afternoon


My system differentiates between these two types of levels, my system doesn't mean the neural network but my interpretation of its signals

What's the tricky part? The bounce from the lower boundary of the channel, it was clear to the horse).


 
mytarmailS:

Well, why not use it?

it is

NS also earns, and oddly enough and loses sometimes

That is, when the signal is produced in reverse, it will be both again

Why would I use it?

 
khorosh:

What's so tricky about it? A bounce from the lower limit of the channel, it was clear to the horse).


)))) omg...

Why didn't you buy it? Are you dumber than a horse?

Alexei Tarabanov:
What is the target?

what's predictable

Renat Akhtyamov:

Of course my answer is why use it?

so there's no reason)


I want you to start generating ideas, and not try to copy mine, I showed the direction and tried to show that there is money in the other side

 
mytarmailS:

I want you to start generating ideas instead of trying to copy mine, I showed the direction and tried to show that there is money in that side

and as always, typical of any neuron, it remains to be added:

in a flat