Machine learning in trading: theory, models, practice and algo-trading - page 1951
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 made a tough decision today - I gave up clustering, because it's d :)
even seasonal patterns are better pulled out by hand than with it (in terms of stability on new data)
Because ensupervised is almost impossible to control, it lives by itself. Can be good for fast data partitioning only
the plan is to play around with umap and see what it is
If anyone's interested, it's a bot-generating thing.
Then in the mountains or at sea. Before the summer is over.
Then to the mountains or to the sea. Before the summer is over
I have an idea for another approach) I don't want to go to the sea yet, there are covids everywhere.
For a network with 1 neuron and 100 inputs. On the left all inputs, in the center the last 10. On the right 10 neurons 100 inputs.
Weights for the grid to intersect with ma(100) 100 inputs on the left, ma(50) 100 inputs on the right.
What information does this give?
today i made a difficult decision - i gave up clustering, because it is d :)
Ooooo!!! How unexpected)) remember someone told me that d)) this, like many times)) clever guy, I do not remember the name...
I'm planning to give the umap a try and see what it's all about.
Just make sure you learn the theory, not even about the algorithm itself, but about dimensionality reduction, what it is, what it's for, how you should use it properly and what you shouldn't, etc...
Otherwise the output will be a rubbish, I practically guarantee it.
Woohoo!!! How unexpected)) remember someone told me that d)) it, like even many times))
Just make sure you study the theory, not even the algorithm itself, but the dimension reduction, what it is, what it's for, how you should use it properly and how you shouldn't, etc...
Otherwise, the output will be a mishmash, which I can practically guarantee.
Until you try it yourself, as they say...
What information does this give?
On the upper charts, you can see that the grid uses the last 1.2 inputs (that is 1.2 previous price values).
On the lower charts, it uses as many prices as necessary to calculate the MA, if MA(50), that is exactly the number of inputs the network uses.
The upper network is poorly trained and has an accuracy of about 60%, the lower one is 99% accurate.
The upper network tries to determine the direction of the next bar based on the previous bar increments (prediction).
The lower net simply determines the direction of price and Ma intersections on the current bar, increments on the input (no prognosis).
The upper graphs show that the grid uses the last 1.2 inputs (that is 1.2 previous price values).
On the lower charts, it uses as many prices as necessary to calculate the MA, if ma(50), that is exactly the number of inputs the network uses.
The upper network is poorly trained, there is about 60% accuracy, the lower one gives 99%.
а. You can look at just
Just make sure you learn the theory, not even about the algorithm itself, but about dimensionality reduction, what it is, what it's for, how you should use it correctly and how you shouldn't, etc...
Something seems to me that it won't help much. This is information compression. If you compress garbage, it will be compressed garbage.
If you add 1 good trash chip to 2500, the algorithm won't notice it much, and its influence on the final result will be if more than 1/2500, then not much. Even if it's 1/100, you won't notice it on the graph.
The only thing I expect is that high-correlated features will sort of merge into one.
look for a bot in the mart top for mt5 that trades seasonal
The name starts with a P
And think about how to reverse. I don't get the same even, but the topic is working
а. You can just look at