Machine learning in trading: theory, models, practice and algo-trading - page 677
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
What was the maximum number of neurons used in your NS? What were the structures of the NS?
I don't remember anymore, up to 100 neurons, max. 2 intrinsic layers
Then I switched to scaffolding and forgot what it means to wait for the NS to learn, because of that I had time to check many variants of TS
MLP and RDF are exactly the same in their capabilities
Once again we are talking about forecasting.
I still don't understand why you're all so obsessed with forecasting? Why?
My NS is purely about classification - just looking for a statistically appropriate moment, favorable conditions for a trade.
Whether the deal will be successful, what kind of profit in the deal (if any) - it is unknown, and all this remains behind the scenes. Like - he knows no fear, he knows no reproach, he knows nothing at all. (с)
Statistically, the NS correctly identifies a good deal 60-80% of the time. The deal simply with a profit >0, including commission, etc. is considered successful.
I've written it for the tenth time, somewhere I've even cited specific figures. Probably I shouldn't do it now).
Once again we are talking about forecasting.
I still don't understand why you're all so obsessed with forecasting? Why?
My NS is purely about classification - just looking for a statistically appropriate moment, favorable conditions for a trade.
Whether the deal will be successful, what kind of profit in the deal (if any) - it is unknown, and all this remains behind the scenes. Like - he knows no fear, he knows no reproach, he knows nothing at all. (с)
Statistically, the NS correctly identifies a good deal 60-80% of the time. The deal simply with a profit >0, including commission, etc. is considered successful.
I've written it for the tenth time, somewhere I've even cited specific figures. Probably, I shouldn't do it now).
Yuri, how come you all do not understand that predict and forecast are one and the same, and you can get probabilities in both cases?
Yuri, how can you all not understand that predict and forecast are the same thing, both can be extracted probabilities
I do not understand). You all want to know the future, what will be the price in time T. And I absolutely do not care what the price will be.
Originally NS was meant to replace logic in usual TS with taught logic, to get rid of endless If-else and < > and thinking of a lot of conditions. And the logic in NS turned out to be much more effective, because it is looking for these set of conditions.
Where is the prediction here? Just statistics, nothing more.
I don't understand). You all want the future to know what the price will be in time T. And I absolutely do not care what the price will be.
Originally the use of NS was conceived as a replacement of logic in usual TS with trained logic, to get rid of endless If-else and < > and thinking about a lot of conditions.
Where is the prediction here? Only statistics, nothing more.
We want to know the difference between the current price and the predicted one.
In the case of regression, these are absolute values, while in the case of classification they are classes.
we want to know the difference between the current price and the predicted one
in the case of regression are absolute values, in the case of classification are classes
I know what you want). I'm sorry, are you interested in future price or successful trades? I'm interested in trades, and I'm only looking for potential trade entries. And the price will go where it will go. And then it is monitored by the management of the deal.
Let's leave it at that. This is not the first time we've talked about this.I know what you want). I'm sorry, are you interested in future price or successful trades? I'm interested in deals, and I'm only looking for potential deal entries. And the price will go where it will go. From there, it is monitored by trade support.
Well, let's leave it at that. This is not the first time we've talked about it.Potential entries into the trades are always forecasting and nothing else.
if nothing is predicted, for example the probability of a certain class, then the NS is not needed at all
before opening each deal the NS issues a forecast, no matter what... class or price
how else to write :)
Potential entries into trades are always forecasting and nothing else
if nothing is predicted, for example the probability of a certain class, then the NS is not needed at all
before opening each deal the NS gives a forecast, no matter what... class or price
how else to write :)
I don't really need NS.) You can do with logic, but NS allows you to do it much more effectively and get along with formal procedures instead of developing complicated logic. NS is only a tool for solving a specific problem, nothing more. This is exactly what Haykin writes about.
And in general, you are talking about the same coin, only you can not believe in the existence of the other side. =)
))) So there is no other side! No! Everyone is talking about it and looking for it, but no one has shown it yet). In general, the big question - is the market predictable in principle? Even on this there is no answer. You can only believe in it or not.
Maxim says everything that my system predicts. No, it merely states the fact that the set of inputs belongs to the A or B classes.
))) So there is no other side! No! Everyone is talking about it and looking for it, but no one has shown it yet). In general, the big question - is the market predictable in principle? Even on this there is no answer. You can only believe in it or not.
Maxim says everything that my system predicts. No, it merely states the fact that the set of inputs belongs to the class A or B. No more than that.
It doesn't know what class the signal should really belong to, because the future is unknown, respectively - it predicts belonging to a class with a probability of 60-80%