Machine learning in trading: theory, models, practice and algo-trading - page 3389

New comment
 
Maxim Dmitrievsky #:
Well, do like the rest of the world. That's the kind of answers you'll get.
You can take 100500 optimisers and smear yourself with them, also an option :)
What the hell are you talking about?

If there is an algorithm that solves your problems, why the hell should you invent your own square wheel with your head held high?
 
Maxim Dmitrievsky #:

I've sketched out some basic theses on kozul, for those who find it difficult to read books in English. And an example in python, how it works best, according to my version. Do you want the article?


Nnnada))))

 
mytarmailS #:
What the hell are you talking about?

If there is an algorithm that solves your problems, why the hell should you invent your own square wheel with your head held high?
Well, give me such an algorithm, which will build profitable TS :) I will call you later from the Maldives, I will thank you. I'll send you a seashell.
 
Maxim Dmitrievsky #:
Well, give me an algorithm that will build profile TCs :)
You can either talk about one thing or about everything and nothing.
L - logic.

We were talking about unstructured data, now we're talking about TCs? Fuck TCs, let's talk about the money printing press and so on..... endlessly about nothing.
 
mytarmailS #:
You can either talk about one thing or about everything and nothing.
L - logic.

We were talking about unstructured data, now we're talking about TC? Fuck TC, let's talk about the money printing machine and then..... endlessly about nothing.
Initially we are talking about the machine, the rest is details :) bag of wards is not very applicable on time series, so there is something wrong with it.
 
Maxim Dmitrievsky #:
Initially talking about the machine tool, the rest are parts :)
A machine is made up of parts, not a machine.
Bag of Wards isn't much use on time series, so there's something wrong with it.
Because Bow is not for time series.

Who says the market is a time series?
Is there any method/algorithm for BP that works on the market? There isn't? I wonder why that is.
 
mytarmailS #:
Because Bow is not for time series.

Who said that the market is a time series?
Can there be a single method/algorithm for BP that works for the market? No? I wonder why that is.
Depends on what market and how it works.
This is a pointless conversation.
 
Maxim Dmitrievsky #:

the same line in the dataset

if you only have 1,000 rows

Roughly speaking, if you have 18+ features, you are training the classifier to remember every row because they don't even repeat themselves

and in causal inference you can't match examples to calculate statistics.

Yes I agree about memorisation - I wrote about it in this thread a long time ago. That's the reason I work with leaves, assessing their value.

Ideally to find a couple of features to describe the whole sample, but nobody has managed to do that yet. On the other hand, reducing the dimensionality will reduce the number of features directly involved in training, but I doubt that this mess will improve the result.

 
mytarmailS #:
1. Any feature values

2. I will surprise you, no one cares how the features were created, everyone evaluates features based on response only

1. So then it's a sample, right?

2. Hmm, I thought this algorithm was trying to evaluate the structure of the rule, like if there are small deviations in splits with other rules, then it's the same thing.

I don't know why you can't explain the process... even if you don't fully understand it yourself, there's no shame in admitting it.

We have a similar concept of working with data in MO, but instead of exchanging ideas, there are constant tensions.

Today I read the thread from NG, and realised that I haven't missed anything....

 
Aleksey Vyazmikin #:

1. So then it's a sample, right?

2. Hmm, I thought this algorithm is trying to estimate the structure of the rule, like if there are small deviations in splits with other rules/leaves, then it's the same thing.

I don't know why you can't explain the process... even if you don't fully understand it yourself, there's no shame in admitting it.

We have a similar concept of working with data in MO, but instead of exchanging ideas, there are constant tensions.

Today I read the thread from NG, and realised that I haven't missed anything....

I think I've answered all the questions
1...338233833384338533863387338833893390339133923393339433953396...3699
New comment