Machine learning in trading: theory, models, practice and algo-trading - page 3646

New comment
 
The random task generator has calmed down, can we go and have tea?
 
Andrey Dik #:

I gave you an example of where:
(a) A stable pattern described by the formula, and so is uniform for any stretch of time.
b) even in the presence of a) there is no guaranteed way to find the regularity.

I've given food for thought, so far, mission accomplished.

Now the claim is about the lack of a universal algorithm in the MoD, suitable for all cases of life. Well, yes, there is no such thing. We have to find a way out. There are no miracles.

For example, if we are talking about a class of problems (very narrow and not particularly meaningful for us), where we need to restore the dependence given by a formula from a graph, we can use methods related to formal grammars.

 
Aleksey Nikolayev #:

Now the claim is that there is no universal algorithm in the MoD, suitable for all cases of life. Well, yes, there is no such thing. You have to find a way. There are no miracles.

Well, finally.)))) Not "now," I was talking about it originally. It's good that you've said it now.
 
Maxim Dmitrievsky #:
The random generator's calmed down, can we go have tea?

That's right, it's time for your "mad tea party" -- "If every man did his own thing, the earth would turn faster." [Alice in Wonderland, Lewis Carroll]

 
Andrey Dik #:

Dear, what are you doing in the MO thread if you think you are dealing with random processes in the market? Then you are trolling, not me. You should create a separate thread on random processes and post there, about matstat and so on.

Machine learning is part of statistics. So it's a revelation to you. And it is MO that is applied all over the world to financial series. This is also a revelation to you.

 
I wonder if we create another MO thread, will Dick come there too?
 
Andrey Dik #:
Well, finally.))) Only it wasn't "now" that I was originally talking about it. It's good that you voiced it yourself now.

You said a lot of things) I remember the most about reducing MO to AO)

This thread wouldn't be so long if everyone believed in the existence of the only correct algorithm. On the contrary, everyone has their own algorithms.

 
Andrey Dik #:

Dear, what are you doing in the MO thread if you think you aredealing with random processes in the market? Then you are trolling, not me. You should create a separate thread on random processes and post there, about matstat and so on.

I just now noticed that Dick thinks the market is NOT random.

Does it make sense to pay attention to his posts?

 
It was stated that D. has a special criterion, thanks to which it is possible to choose those weights of NS, which will be the only true and will predict the function. Then it was stated that it is impossible to predict it. I predict it without any special criteria.

Question for the blabbermouth: any results to follow from his side? I'm just kind of looking forward to the 10 minutes.

Attention, connoisseurs. the correct answer is.

Either that or you're out.
 
Question for MO experts:

Each time Python finishes training, the error has a unique number.

Each time training ends with a unique set of weights.



If we loop the training with constant reinitialisation of weights, is there any chance that after some time the error will fly to the bottom on both the training sample and the test sample?
1...363936403641364236433644364536463647364836493650365136523653...3741
New comment