Machine learning in trading: theory, models, practice and algo-trading - page 682
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
I ) We take a neuron, make it trade in the live mode, and at the same time select its configuration. It drives the dealer crazy with its profits, he starts to move the price against it, but it is somehow ready for that and still trades on the upside, the dealer copies the trades to the interbank market to make money, it drives the exchange robots crazy, they become stupid and the global market goes into the abyss. The agent interacts with the environment. It's backed trading. I think on a recent Monday some Google was testing their new trading robots with reinforcement, this fits right in perfectly.
It seems to me that reinforcements should still be based on historical data. Or you want to trade robots with random weights?
Learning without a teacher and learning with reinforcement are different things.
The simplest representation of teacherless learning is the autoencoder, the simplest representation of reinforcement learning is the arena where intelligent agents fight for a limited resource.
Or do you want to start trading robots with random weights?
I don't want anything with the backing now, I just gave an example of what the picture says.
I don't know how to initialize the scales correctly. But for example there is a package rneat, there first set of neurons is completely random, 99% of them will sit in the same transaction all day. Then in a cycle the genetics will create a new set of neurons based on past most successful ones, for some number of such cycles most of neurons should become more adequate.
Gentlemen! Can someone give me an example of of a real deal, based on a prediction. NS? Even if unsuccessful, but with a full description - how many inputs, what inputs, what outputs, forecast depth, etc.
There are now cloud services for machine learning with a visual development environment. You can use ML there for free, without going into too much detail.
You just need to have the data to train. You have plenty of ticks, so no problem with that.
Try this one:
https://studio.azureml.net
Didn't you look at the previous page?)))#6809
Sorry, Wizard... On the way to the Grail I began to fuss something...
Aleksey Terentev - thank you!
Good afternoon to all! I am a newcomer to this site and to the trend market. I'm studying at the university. I got my diploma in Technical Indicators: The Directional Movement System. Is it possible to connect this theme to the machine learning and what else I can add to it? I would like to thank you in advance.
The diploma is usually written after 4-6 years of study (in the profile in which you studied). You are not a beginner in this field, and should be more literate than many people here. Or you have studied something else, and the diploma on a theme of markets was given (in what, as you write - the beginner)?
It is unlikely to be related to the MO technical indicators.
I don't know how to make a new one, but you can at least prove that these indicators are outdated now.
You can try to predict the direction of the next bar based on these indicators.
Take eurusd h1 in a couple of months for example. Put indicator parameters and MO model parameters into the genetics. In the fitness function of genetics - train a model with K-fold crossvalidation. If the genetics finds suitable parameters of both the model and indicators for successful trading - it will be a miracle, but most likely not.
And then the same thing but for rice prices by time a dozen years before those indicators were created. Probably the indicator will even work on that data. It will be interesting to see year by year how the indicator started to weaken before it rolled off.
Vizard_ , don't give me that. SanSanych is one of the few who just takes it and explains how not to do it. Unlike many who simply sow misinformation and incorrect methods.