Machine learning in trading: theory, models, practice and algo-trading - page 2194
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
And we all continue to admire))))
Why did you have to make such a big deal out of it, I ask you....
And we all continue to admire))))
What was the point of showing off, I ask you....
The circus has left, but the clowns stayed.
Interesting example of sampling new points from CVAE, an example of range selection for sampling (in the feature space). Transformation of the feature space. Visualization of classes/distributions.
if you have one in python, that's greatHere is the formulation of the variational autocoder , and its revision in keras with creation of its loss function. I'll be poking around tonight)
https://wiseodd.github.io/techblog/2016/12/10/variational-autoencoder/
Here is the formulation of the variational autocoder, and its revision in keras with the creation of its loss function. I'll be poking around tonight.)
https://wiseodd.github.io/techblog/2016/12/10/variational-autoencoder/
Yeah, cool. My tensor for some reason does not want to get up, so I'll look at the torch
So far I'm not even so much interested in the implementation, as in the approach itself. How to set distributions, interpret, etc. Because it is possible to do in different ways.
here are interesting articles 1, 2, give an intuition for what's going on inside
Anyway, I did it... I'm not too happy with the result... I was better at predicting the trend lines
it is interesting that the price is not hanging out in the predicted channel so much as bouncing off it when it breaks
pictures... everything after the vertical line is a forecast
I put the date set on the previous page
Anyway, I did it... I'm not too happy with the result... I was better at predicting the trend lines
it is interesting that the price is not hanging out in the predicted channel so much as bouncing off it when it breaks
pictures... everything after the vertical line is a forecast
the date set is on the previous page
I was forecasting distributions n-bars ahead (mean and variance). Depending on the distribution I chose a strategy. Learned well, on new data poorly.
But with the new resampling approach, it might work
He's been stomping around for four years. During that time, he could have made 10K bucks on the construction site. Let's see what happens. He won't blow it. He'll be on the plus side.
I predicted distributions by n-bars in front (mean and variance). Depending on the distribution I chose a strategy. Trained well, on new data poorly.
But with the new resampling approach, maybe it will work.
Yes... this is not an easy thing... I have one more idea, if it does not work, I do not know what to try((
Yes ... this is not an easy thing ... there is one more idea, if it does not work, I do not know what to try((
Extremes on both sides)))) data seems to be all that can be counted on the plot made. Now it is more difficult, to divide the section into zones of states manually.... I'm too shy to do it.... It's complicated)))))
Extremes on both sides)))) data seems to be all that can be counted on the section made. Now it is more difficult, to divide the section into zones of states manually.... I'm too scared to do it.... Complicated)))))
Yes, any attempt to predict levels is a hundred times harder than the prediction of the trace value, everything is not clear, I have to invent a lot of new and not trivial to implement the idea...
But it's interesting, and I have a rosy dream it will work.)