Machine learning in trading: theory, models, practice and algo-trading - page 1815
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
it's not naive. it's just for fun....)))) Here about the ns predictors of mo model chips and there are no formulas here)))))) in addition to that the secret grail wonders))))) There are no secrets in mat models, you just have to understand them.))))))
What? It's OK. As the Poles say, all your secrets are studied, nothing new unfortunately.
Yes gentlemen, you can teach NS. But it will work well on history (that's for sure) or only on exceptionally ideal conditions, which the market will try not to provide you in the future.
I do not know how you can better predict the behavior of the price.
I hope you will understand that this is the current moment.
Yes gentlemen, you can teach NS. But it will work well on history (that's for sure) or only on exceptionally ideal conditions, which the market will try not to provide you in the future.
NS is a multivariate optimization!
What do you want from the optimization? That it flew into space, and before that it built a spaceship by itself?
If a man can't find a pattern in the graph by himself, what do you want from the NS?
The NS has an average of 100 to 1000 neurons, andthe humanbrain has 86 billionneurons, do you feel the difference?
I'll put it simply, you confuse/compare the usual optimization with artificial intelligence, hence all your frustrations, the problem is in you and not in the NS! (I don't mean you personally, most people here think so)
NS already knows a lot, they do amazing things ...
Karoch. summarizing, dry balance - if you have a system that gives 70% positive inputs for example the NS will help to make 90-95% of the positive inputs, if your system gives 50/50 then with the NS it will give 45/55 and moreover the positive inputs will be 45% !
Yes, it's unpleasant, it's useless... What do you want from optimization? With garbage at the input? A miracle? Well, everybody wants so here, the result is obvious...
The NS is multidimensional optimization!
What do you want from the optimization? that she flew into space, and before that she built a spaceship herself?
If a person can not find a pattern in the graph by himself, what do you want from the NS?
The NS has an average of 100 to 1000 neurons, andthe humanbrain has 86 billionneurons, do you feel the difference?
I'll put it simply, you confuse/compare the usual optimization with artificial intelligence, hence all your frustrations, the problem is in you and not in the NS! (I don't mean you personally, most people here think so)
NS already knows a lot, they do amazing things ...
Karoch. summarizing, dry balance - if you have a system that gives 70% positive inputs for example the NS will help to make 90-95% of the positive inputs, if your system gives 50/50 then with the NS it will give 45/55 and moreover the positive inputs will be 45% !
Yes, it's unpleasant, it's useless... And what do you want from optimization? With garbage on the input? A miracle? Well, everybody wants so here, the output is obvious...
I'm already glad that at least one person is ready to talk.
What is the point of all this? If you load garbage into the NS, what's the expected output? You can probably guess.
And so all beautiful posts with even a scientific view.
It's nice to have at least one person to talk to.
What is the point of all this? If you load garbage into the NS, what is expected to come out? You can probably guess.
And so all beautiful posts with even a scientific view.
I reread your post again and carefully.
I'll put it simply, you confuse/compare the usual optimization with artificial intelligence, hence all your frustrations, the problem is in you and not in the NS!
I think it makes no sense to optimize.
You can only optimize the story, but it is already far from reality.
There are a lot of hints for the future and only they should be used.
That is where the AI is needed, not for the story)))
Yes gentlemen, you can teach NS. But it will work well on history (that's for sure) or only on exceptionally ideal conditions, which the market will try not to provide you in the future.
I do not know how you can better predict the behavior of the price.
I hope you can understand that this is the current moment.
Where does the stable belief that the price movement is absolute chaos come from? There is such a profession as "trader", trader uses regularities that are there and thus earns his bread. A neural network also finds these regularities. Yes, it is a non-trivial task, but it is solvable and there are examples.
Another all-spreader, it works and predicts normally. It is not perfect, of course, but it is on the level of an average trader.
Where is the stable belief that the price movement is absolute chaos? There is such a profession as "trader", trader uses regularities existing there and thus earns his bread. A neural network also finds these regularities. Yes, it is a non-trivial task, but it is solvable and there are examples.
Okay, I'll go get a beer, then I'll answer it.)
Yes gentlemen, you can teach NS. But it will work well on history (that's for sure) or only in extremely ideal conditions, which the market will try not to provide you in the future.
The nature of neural networks is such that if it is well trained on a long history, it works well in different conditions, and if the conditions change significantly, then it simply stops predicting and waits until the market becomes familiar to it again. This is the great advantage of neural networks over "human" strategies.
I reread your post again and carefully.
Let me put it simply, you confuse/compare conventional optimization with artificial intelligence, hence all your discrepancies, the problem is in you and not in the NS !!!
I think it makes no sense to optimize.
You can only optimize the story, but it is already far from reality.
There are a lot of hints for the future and only they should be used.
That's where AI is needed, not on the story)))
Please don't take the word optimization as a synonym - adjusted period at the stochastic.
When you analyze your patterns and try to identify quality entry points of say a buy, sell and garbage patterns...
What do you do? You minimize class error(this is optimization), and there are three classes: "buy" , "sale" This is aclassification problem in machine learning.
When you broke down market situations into patterns and gave them numbers, that's the clustering task in machine learning.
When you find the optimal number of patterns, that's optimization as well.
You see, you have been doing machine learning for a long time without even knowing it, only with non-optimal tools, but that's empty.
There are a lot of hints for the future and only they should be used.
That's where AI is needed, not for history))).
You are contradicting yourself.
How do you know that it's a hint at the future? Probably because you've seen it before, and before it was like a story.)
The nature of a neural network is such that if it is well-trained on a long history, it works well in different conditions, and if the conditions change a lot, it just stops predicting and waits until the market becomes familiar to it again. This is a great advantage of neural networks over "human" strategies.
Well the history may change at any time, and the neural network will not change the nature of price movement. Don't even dream). And you can wait for years. Hee-hee