Machine learning in trading: theory, models, practice and algo-trading - page 3673

[Deleted]  

Of the trending ones, gold/silver are better to learn. I haven't tested indices/crypto yet.

Only it looks like gold may be at a peak before a long correction and the ts will still break :)

[Deleted]  
fxsaber #:

Some in the classics (not MO) put conditions in the optimisation criterion such that long positions are considered either always unprofitable or their profit is penalised as a function of duration.


Also for variants with a small number of trades OnTester = 0. And other logical tricks to prevent swans from flying in, etc. I have a lot of input-parameters in the optimisation criteria. Due to their settings, I can search for different types of regularities. Including them in Optimisation is a bad solution, because a "better" pattern will be found.

what is the average lifetime of such ts?

 
Maxim Dmitrievsky #:

what is the average lifespan of these tc's?

We were talking about general principles of selection, not about specific vehicles.


This is a common situation with TCs.


The TS was on the reals - it started to pour. I thought it was broken. I switched it off. Two years later, I accidentally looked at what it showed over the past two years - it was grading. I put it back on the real - it doesn't seem to be hailing. You think whether to switch it off or not.


That's why it is somehow weak to determine the life span.

[Deleted]  
fxsaber #:

It was about general selection principles, not specific TCs.


This is a common situation in TCs.


The TC was on the reals - it started to pour. I thought it was broken. I switched it off. Two years later, I accidentally looked at what it had shown over the past two years - it was hailing. Put it back on the real - it doesn't seem to be hailing. You think whether to switch it off or not.


That's why the definition of the life span is somehow weak.

It turns out that the factory is more stable )
[Deleted]  

Interesting way of clustering to find patterns.

Better save the pdf at once. They like to make access paid if there are a lot of views :)

That is already practically cover the topic of clustering from all sides, as it was before with classification. For full understanding.

 
Maxim Dmitrievsky #:

An interesting way of clustering to find patterns.

Better save the pdf at once. They like to make access paid if there are a lot of views :)

That is already practically cover the topic of clustering from all sides, as it was before with classification. For full understanding.

Good article. The power is growing and it gives results. Already not one model but several and range/cluster of their actions and selection of rules and cluster boundaries also not in the head. And in enough detail))) but the rules-indicators here are not arbitrary, they are set initially, even with a range of parameters. And it is a matter of time to expand the number of indicators or make it arbitrary or indicators-rules genetically sculpted.
 
Maxim Dmitrievsky #:

An interesting way of clustering to find patterns.

Better save the pdf at once. They like to make access paid if there are a lot of views :)

That is already practically cover the topic of clustering from all sides, as it was before with classification. For full understanding.

What's the benefit to us with the following quotes below?

( 2) provide stock trading guidance using a combination of generated trading rules and Bayesian decision making with minimum error rate under the assumption of a multivariate normal probability model;

Technical analysis is based on the assumption that there are stable patterns of behaviour that are time-independent, related to the stock price and will repeat in the future; thus, these patterns can be used for forecasting.


Nothing but outright stupidity can grow out of technical analysis.

[Deleted]  
СанСаныч Фоменко #:

What good is it to us, given the following quotes?

( 2) provide stock trading guidance using a combination of generated trading rules and a Bayesian decision-making method with minimal error rate assuming a multivariate normal probability model;

Technical analysis is based on the assumption that there are stable patterns of behaviour that are time-independent, related to the stock price and will repeat in the future; thus, these patterns can be used for forecasting.


Nothing but outright stupidity can grow out of technical analysis.

Simultaneous clustering by rows and columns.

 
СанСаныч Фоменко #:

What good is it to us, given the following quotes?

( 2) provide stock trading guidance using a combination of generated trading rules and a Bayesian decision-making method with minimal error rate assuming a multivariate normal probability model;

Technical analysis is based on the assumption that there are stable patterns of behaviour that are time-independent, related to the stock price and will repeat in the future; thus, these patterns can be used for forecasting.


Nothing but outright stupidity can grow out of technical analysis.

Not the obvious rules.
[Deleted]  
I don't pay attention to the word-pullers :)