Machine learning in trading: theory, models, practice and algo-trading - page 3303

New comment
 
fxsaber #:
I think it is more complicated. For example, there is an Expert Advisor in the Market that shows different results on real ticks and tester-generated ticks. I can't get into the reasons.

I misled by this adjective.

On the generated ticks - a grail, on real ticks - a plum.

 
fxsaber #:

Misleading with that adjective.

On generated ticks - grail, on real ticks - plum.

Well, and in real trading it will be a drain, because the conditions are even tougher on it

The market can't be changed. They will always buy pictures, not normal TS :)
 
Maxim Dmitrievsky #:

And in real life, it will be a drain, because the conditions are even tougher.

The market can not change. They will always buy pictures, not normal TC :)

That's not the point. The original question was about plum when adding spread. And here it is even more complicated - the spread is there and graalit, but it is generated.

 
fxsaber #:

That's not the point. The original question was about draining when adding spread. And here it is even more complicated - the spread is there and graalit, but it is generated.

How do you know it's graalit? The demo monitoring is not so good. But the pictures from the toaster are beautiful.

It's a regular tick grail, they work on the OOS too.
 
Maxim Dmitrievsky #:

regular teak grail, they work on the OOS too

The backtest trading history has been watched. Bygones.

 

Don't you get bored ... ?

Any broker, or trader with a large deposit, will crush any of your neural networks!

 
Sergey Chalyshev #:

Don't you get bored ... ?

Any broker, or trader with a large deposit, will crush any of your neural networks!

Yeah, in forex.)

 
fxsaber #:

Backtest's trading history. Bygones.

I figured it out, it was the peculiarities of autopartitioning on smaller tf.

 
fxsaber #:

That's not the point. The original question was about draining when adding spread. And here it is even more complicated - the spread is there and graalit, but it is generated.

I'm surprised you're surprised.

About 10 years ago, when I developed my first robot in MQL5, I made millions on the tester. But since I thought it was unbelievable, I started looking for what was wrong.

I didn't know then that "Every Tick" is not real ticks but generated ticks. This site has an algorithm and schemes for generating these ticks.

At that time brokers did not collect tick values yet. And I did it myself. I collected real ticks and stored them in files in portions for about 6 months. I applied them on the tester and got a completely different picture.

Spread has nothing to do with it, if it is not HFT trading or scalper.

While optimising the settings, it is not difficult to find such a combination of parameters, when the robot starts to work synchronously with the generated ticks.

I.e. it catches the pattern of ticks generation, like here:

I think everyone knows about it.


 
MetaQuotes #:
The articleClassification models of Scikit-learn library and their export to ONNX has been published

Here is a list of supported ones from the site itself

https://onnx.ai/sklearn-onnx/supported.html

Thanks for the ready examples in the article.

Supported scikit-learn Models#
  • onnx.ai
, # This version of the operator has been available since version 13. Absolute takes one input data (Tensor ) and produces one output data (Tensor ) where absolute value, y = abs(x), is applied to the tensor elementwise. T tensor(uint8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(int8), tensor(int16), tensor(int32), tensor(int64...
1...329632973298329933003301330233033304330533063307330833093310...3699
New comment