Machine learning in trading: theory, models, practice and algo-trading - page 3203

 
mytarmailS #:

Why ours? Yours.

You've been overheating too much lately, you and Sanych, take a holiday.
 
Maxim Dmitrievsky #:
You've been overheating too much lately, you and Sanych, take a holiday.

Thanks, I'll definitely consider your suggestion in the order of the day.

 

Lira has actually been in OnlyClose mode for years. It is more useful to apply MO to the neighbour's cardiogram.


ZY A logic problem.

There is a certain country where anything can be bought only in lira. And absolutely everything (even salary) has a price tag in U.S. dollars, which in lira depends on its exchange rate.

And now the lira has become three times cheaper against the U.S. dollar. Was it even theoretically possible (for a foreigner or a local) to make money on this?

Let me say up front that we are not talking about Turkey, but about a certain ideal model, which should eventually show exactly what earnings are made on in real time. Turkey included.

 
fxsaber #:
Lira has actually been in OnlyClose mode for years. It's more useful to apply MO to your neighbour's cardiogram.
Practising on cats. I just saw that if there is a regular behaviour, the algorithm finds it too, it does not retrain for noise.
 
Maxim Dmitrievsky #:
We're practising on cats. I just saw that if there is a regular behaviour, the algorithm finds it too, it doesn't retrain for noise.

If the naked eye sees it, the MO will find it, of course. The task of the MO is different - to find a pattern where it is not visible.

 
fxsaber #:

If the naked eye has seen it, the MO will certainly find it. The task of IO is different - to find a pattern where it is not visible.

If the architecture of the MO allows concentrating on certain things, otherwise it will not find it
And to do that, you have to apply expert knowledge. So it's more likely that IO can't find what's hidden.
 
fxsaber #:

If the naked eye has seen it, the MO will certainly find it. The task of the IO is different - to find a pattern where it is not visible.

Tell me, please, how does regularity differ from rote learning from the position of MO?

For example, I have an idea in my head how a pattern of 5 candlesticks should look like. I mark them on the chart, run the price export script, feed it to the neural network, it learns how the patterns should look like and how they should not (the labels of the other sets of prices are different), and then my neural network marks these patterns on the chart. That is, I don't need to write "if, then", the neural network learnt and understood it by itself. I pointed out to it with my finger what and how.

But here is the task "find a pattern", what is it, how does it differ from the example with rote learning?

Maybe finding a pattern is too complicated a task and the neural network at the moment just keeps "rote learning" in proportion to the computer's power.

 
Ivan Butko #:

Tell me, please, how does patterning differ from rote learning?

For example, I have an idea in my head how a pattern of 5 candlesticks should look like. I mark them on the chart, run the price export script, feed it to the neural network, it learns how the patterns should look like and how they should not (the labels of the other sets of prices are different), and then my neural network marks these patterns on the chart. That is, I don't need to write "if, then", the neural network learnt and understood it by itself. I pointed out to it with my finger what and how.

But here is the task "find a pattern", what is it, how does it differ from the example with rote learning?

Maybe finding a pattern is too complicated a task and the neural network at the moment just keeps "rote learning" in proportion to the computer's power.

Two circumstances:

1. You can mark anything. The problem in the future: with what error your MO will find what you have marked.

2. Your "marking" has nothing to do with the MO, the meaning of which is different, for example: we form "teachers", i.e. we mark the areas of the kotir, which are guaranteed to have, for example, logs and shorts. Then we give the MO algorithm this "teacher" and a certain set of input data, for example, indicator values. The MO algorithm forms some sets of indicator values that correspond to the teacher's values. These are the patterns. In principle, these patterns are nothing to us, but the most surprising thing is that in the future these patterns are repeated, though with some error.

 
СанСаныч Фоменко #:

Two circumstances:

1. You can mark anything. The problem in the future is: with what error your MoD will find what you have marked.

2. Your "marking" has nothing to do with the MO, the meaning of which is different, for example: we form a "teacher", i.e. we mark the areas of the kotir, which are guaranteed to have, for example, logs and shorts. Then we give the MO algorithm this "teacher" and a certain set of input data, for example, indicator values. The MO algorithm forms some sets of indicator values that correspond to the teacher's values. These are the patterns. In principle, these patterns are nothing to us, but the most surprising thing is that in the future these patterns are repeated with some error.

Thank you for your reply
 
fxsaber #:

Lira has actually been in OnlyClose mode for years. It is more useful to apply MO to your neighbour's cardiogram.


ZY Logic problem.

There is a certain country where anything can be bought only in lira. At the same time absolutely everything (even salary) has a price tag in U.S. dollars, which in lira depends on its exchange rate.

And now the lira has become three times cheaper against the U.S. dollar. Was it even theoretically possible for a foreigner or a local to make money on this?

Let me say up front that we are not talking about Turkey, but about a certain ideal model, which should eventually show exactly what earnings are made on in real time. Turkey included.

It's not clear whether to earn in lira or in U.S. dollars? Everyone earned in lira as everyone had three times as much)

It is possible to earn in U.S. dollars, for example, if it is possible to take and, most importantly, to repay loans in lira, but most likely, it is impossible according to the terms of the problem. Then it turns out that lira is not money in this country.