Machine learning in trading: theory, models, practice and algo-trading - page 3189
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
Well, and another not quite scientific addition - the indicator collected in the sample must somehow correlate with profit. The difference from the SB on the basis of a completely left-wing characteristic makes no sense, IMHO.
Well and simplification of the calculation scheme, in addition to speeding up, will give greater confidence in the absence of errors and, accordingly, a lower probability of self-deception.
Come up with a simplified calculation scheme for simulations.
For some (not absolute) confidence in meaningfulness, the result for real data should fall at least in the 5% tail of the sample (left or right). But the sample should be several thousand at least.
If I change the conditions of the experiment, to the following:
1. On the original sample we find quantum segments, which are supposed to be used in the future, as a result one quantum table is formed - further we work only with it.
2. Randomly generate a target one with the same parameters as the original - 1000 cycles.
3. Count how many quantum segments are selected in comparison with the original variant. They can be the same or less.
4. Evaluate through the standard deviation. If the spread is small, then random targets have a big chance to get into the selected quantum segments.
What do you think?
Well, and another not quite scientific addition - the indicator collected in the sample must somehow correlate with profit. The distinction from the SB on the basis of a completely left-wing characteristic makes no sense, IMHO.
I.e. change the target so that the profit would be comparable to the original? I'm not quite sure what was meant.
That's the finish line. What's there to criticise?) Did you realise what you did and why? If you did, why did you ask on the forum?)
Now it's about you not understanding what I'm doing, and I can see that as I'm evaluating feedback on questions and assertions on your part.
When you figure it out, we'll get back to dialogue.
Now it's about you not understanding what I'm doing, and it's obvious to me as I evaluate feedback on questions and assertions on your part.
When you figure it out, let's get back to the dialogue.
If I change the conditions of the experiment to the following:
1. On the original sample we find quantum segments, which are supposed to be further used, as a result one quantum table is formed - further we work only with it.
2. Randomly generate a target one with the same parameters as the original - 1000 cycles.
3. Count how many quantum segments are selected in comparison with the original variant. They can be the same or less.
4. Estimate through the standard deviation. If the spread is small, random targets have a big chance to get into the selected quantum segments.
What do you think?
Relate it somehow to the profit, at least approximately, and compare the real profit with a sample of random profits. Check that there are no errors is equality of the average profit on the sample to zero. Check the significance of the positivity of the real profit relative to the sample - the three sigma rule.
I am not ready to go into details of your task, as my own tasks are too busy.
I.e. change the target so that the profit is comparable to the original? I'm not sure what was meant.
Are your quants designed to make a profit? Is there a scheme for that? Make it extremely simplistic so that you can calculate, even if roughly, a quick sample and check if the real result hits the tail of that sample.
Your willingness to require people to delve into your mindset, accompanied by your complete unwillingness to delve into simple and widely known ideas like Monte Carlo, is tiresome.
I think I've had enough.
Your willingness to demand that people get into your mindset is tiresome, accompanied by your complete unwillingness to get into simple and widely known ideas like Monte Carlo.
I think I've had enough.
You're looking for sequences that are different from the sequences on sb. You won't find any.
About the strict sequence I wrote just as an example for clarity. And, I wrote that the solution of this problem can improve the stability of the model. But the solution can be different.
Even without solving the above mentioned problem - selection of the correct quantum table improves learning, which was tested by me on dozens of samples.
Then I showed how you can quickly do preprocessing for training, cleaning the sample from inconsistent data. You can see on the gifs that you can even get a profitable model on new data with this method.
All in all, the approach works, its development is my goal.
So to say it doesn't is to deny reality.
I don't believe that price is in pure SB form, the nature of which can't be parsed at least partially. If it is pure SB, then the whole branch is a mistake.
Relate it somehow to the profit, at least approximately, and compare the real profit with a sample of random profits. The check that there are no errors is that the average profit on the sample is equal to zero. Check the significance of the positivity of the real profit relative to the sample - the rule of three sigma.
I am not ready to go into details of your task, as my own tasks are too busy.
What does profit have to do with it, when we are talking about preprocessing data for further classification?
Are your quanta designed for profit extraction? Is there some scheme for this? Make its extreme simplification to calculate even approximately but quickly a sample and check if the real result falls into the tail of this sample.
Your willingness to require people to delve into your mindset, accompanied by your complete unwillingness to delve into simple and widely known ideas like Monte Carlo, is tiresome.
I think I've had enough.
Everyone has the right to manage their own time.
But, apparently, you did not understand the question on which you gave advice.
Thank you for trying to help.