Pair theory - page 3

New comment
 
Also - imho, you don't have to believe in methods. I'll repeat a tired question for you: Do you use forecasting in auto-trading?
 
Svinozavr:

Their own/you. ))) Shit. Well, if you're following, you know the general outline.


It's a puzzle, I can't figure out the general contours of it from your posts. What is it based on?
 
tara:
Also - imho, it is not necessary to believe in methods. Let me repeat a question that has become very old for you: Do you use forecasting in auto-trading?


I do not use autotrading - my TS cannot be automated in MT.

If one trades, one way or another, one makes a prediction about the future. Or a guess as to the dynamics of the price.

 
FAGOTT:


I do not use autotrade - my TS cannot be automated in MT.

If one trades, one way or another one makes a prediction about the future. Or an assumption about price dynamics.

OR
 

prediction or conjecture is sophistry

What difference does it make in principle? Are you predicting that the price will rise or are you predicting that the price will rise?

 
FAGOTT: Adapt this system for forex and run the story for three years. And they will need thousands of times less computing power - the only indicators they will need are price of the asset and volume, for example on one-minute bars.

Banks have enough money for supercomputers.

This idea has already been described on the forum.

You underestimate the complexity of the market and its fundamental difference from the weather phenomena. Weather is described using known difurcas, and the main problem in the accuracy of its prediction is not the lack of a model, but the accuracy and density of the experimental data. The main influencing factors have long been known.

For the market - at least for one pair - I am not aware of these factors. And it is unlikely that anyone is more or less fully aware of them.

If you think that all patterns can be found with nerve grids, you are mistaken.

 

maybe.

But on the other hand, there are, say, 1000 indicators, oscillators etc. in TA. They have all been automated and run through history in various variations and combinations for a long time. But there is still no grail.

Probably to search and search indeed, but so far the TA is of little use.

 
FAGOTT:

prediction or conjecture is sophistry

What difference does it make in principle? Are you predicting that the price will rise or are you predicting that it will rise?

If you confront the concepts, you're right. It is sophistry indeed.

But that's not what we're talking about.

 
FAGOTT:

maybe.

But on the other hand there are, say, 1000 indicators, oscillators, etc. in TA. They have all been automated and run through history in various variations and combinations for a long time. But there is still no grail.

Sadly, they all share the same basis - price.
 

The hypothesis of existence of pairs and antipairs is quite logical and easy to test.

Such pairs of FI (financial instruments) must show high absolute QC value (correlation coefficient), or rather QC must be high at each point of interval, where pairs behave in the same direction. The sample for TK calculations can be different. Here is an example:

The thin green line is the average QC value (on a sample of 24 hours) at each point in time over the last 2 months. The sharp step corresponds to the time of news release in USA.

Now the same, but the QC sample is reduced from a day to 6 hours:

Lastly, the sample for calculating QC is one hour:

The intervals you are looking for are where the green line is high. The higher the better: the more co-directional the pairs will behave on average.

So how do you find these places and pairs? Again it is not difficult. It is necessary to combine two approaches: the first and the second. The result is a table, and a simple analysis will give you the answer.

1234567
New comment