Is there a pattern to the chaos? Let's try to find it! Machine learning on the example of a specific sample. - page 29
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
Well, this is purely your system, and it has nothing to do with the data I gave, because you didn't use any other data for analysis, did you?
I am attaching a file - please apply on it the model that you have previously trained - I am interested in the result.
PR=183856 +trades=693 -trades=18
At spread=0 and commission=0.
So your genetics is responsible for the data fed into the inputs to the network? And the data itself is the time series bias?
I already wrote the answer to the first question ... :) No bias.
Regards, RomFil.
Also for different parts of the graphs we need different depths of samples fed to the input of neural networks. That is, neural networks with different sampling depths have different accuracy in different parts of the graph. So, the "right" committee allows to respond correctly on the whole length of samples. And especially that this committee itself determines this correctness. Perhaps this is already the rudiments of AI ... :)
Interesting.
I myself train and compare on 5,10,20,50 thousand rows and everyone trades differently with different results. Combining them together is an interesting idea. Do you average?
Usually if you average the different trading models, they contradict each other and start to trade less often, only when the majority agree.
How can 5-10 models determine correctness by themselves? Do you mean average?
PR=183856 +trades=693 -trades=18
At spread=0 and commission=0.
Try to apply the model on this data and I won't torture you with it anymore :)
Well, this is purely your system, and it has nothing to do with the data I gave, because you didn't use any other data for analysis, did you?
I am attaching a file - please apply on it the model that you have previously trained - I am interested in the result.
As it turned out, it doesn't matter what data ...
Actually the graph formed by this algorithm
look like this (and at each run a different graph is obtained for known reasons):
Traine sample first 10000 values, the remaining 2000 is a test.
The result is this:
PR=406206 +trades=299 -trades=34
This is the end of the story. All the best and dreams come true.
Regards, RomFil.
Try applying the model on this data again and I won't torture you with it anymore :)
PR=116823 +trades=977 -trades=16
The answer to the first question has already been written ... :) No offset.
Regards, RomFil.
You yourself write " Yes, almost pure values, different depths, different windows, etc. ".
The result is this:
PR=406206 +trades=299 -trades=34
There are 2000 signals on the chart, but you have 333 in the description - or I don't understand something again....
Okay, if this is the chart of the last sample, it turns out that the model trained on EURUSD works perfectly on 3 different currency instruments, including cross. I guess it's time for a Nobel Prize!
This is the end of the story. All the best and dreams come true.
Regards, RomFil.
Thank you for an interesting evening, and all the best to you!
PR=116823 +trades=977 -trades=16
shock, shock.
Actually, the graph generated by this algorithm look like this (and at each launch a different graph is obtained for well-known reasons):
On random graph and training on it?
Interesting.
I myself teach and compare on 5,10,20,50k lines and everyone trades differently with different results. Combining them together is an interesting idea. Do you average?
Usually if you average trading models, they contradict each other and start to trade less often, only when most of them agree.
How can 5-10 models determine correctness by themselves? Do you mean average?
You ask the right questions!!! :)
But I won't reveal this secret. :) But I will say one thing, that the output of the committee itself determines the "correctness" of this or that network. No averaging.