From theory to practice - page 651

 
Renat Akhtyamov:

I explained the calculation period above - it does not exist, as a quotire is a non-periodic series

in which a quote is allowed to have practically negligible increments, with sufficiently wide possibilities for price management

is proved by Fourier transforms.

I would argue about a 2-3% overweight, but I won't.

Just to give you an example - black swan on the pound in '16, about 90% stood to buy....

And the funny thing is that it was a no-break for all markets, huge livantos

Can a counter-trend strategy withstand such a move? The answer is hopefully obvious.

And what's above is a reversal strategy

The author is off topic, he's right.

I meant globally, I've been taking the whole story for the last 15 years. As I have no idea how to calculate the optimum calculation period.
 
Martin Cheguevara:
...
1 option to solve price corridors based on countering their own highs and lows. This requires getting rid of the impact of sharp price spikes.
2. The solution variant is finding the asymmetry of two forces. The problem here is in the calculation period that should also be adaptive to avoid being too sensitive or vice versa. So far there is no solution to this problem.

I give an indicator on the 1st item, for "it's all rock'n'roll" ;) (ok, but only for dabbling)

On the second - let's say everything is on sale, will there be asymmetry? The answer is YES.

Files:
i-Regr.2.mq4  22 kb
 
Renat Akhtyamov:

I give an indicator on the 1st point, because "it's all rock'n'roll" ;) (ok, but only for dabbling)

on the second - let's say everything is on sale, will there be asymmetry? The answer is yes.

I can give you a hundred such indicators, both regression-based and spline-based and whatever. What is the point? Rock'n'roll or not, it works unlike the second option.
What is on sale based on which instrument? These are reasonable questions that need to be clarified, you probably understand.
OK... we're getting lyrical again)
The essence of our conversations will not change anyway ...
 
Martin Cheguevara:
I can give you a hundred indicators based on regression and on splines, or on anything else. What is the point? You may or may not rock and roll, but it works unlike the second option.
What is on sale based on what you have decided and on what instrument? These are reasonable questions that need to be clarified, you probably realise.

It's not working.

Move on

 
Well ... I hope the few people who know who I mean and who really understand what I mean understood. That's enough for me:).
If there will be decisions about adaptive asymmetry analysis, write a separate thread, as it is really just a fundamental thing, which will be effective not only in forex.
Who missed the discussion, see #6498 there everything you need to know about this branch. The question seems to be exhausted at this point in time.
I am ready to share with people who know my workings which give results.
But this is unrealistic because branches quickly littered grails opinions guesses forecasts and so on.
 

I listened and listened to CheGevara (about Gauss on large TFs and quantile=1.6, etc.) and decided to make a simple indicator:

GBPJPY 2018 pair.

Used:

1. CLOSE M1

2. sliding window - week (7200 values CLOSE M1)

3. simple MA

4. dispersion calculation according to the formula from this branch + quantile =1.6

Looking at:

A total of 7 trades would have been made in 2018 (+6/-1)

Total profit: +733 full points.

If anyone is sure this is the Grail, let them set up a TS and post it here. Right to the branch. Let the suffering people use.

 
Alexander_K2:

I listened and listened to CheGevara (about Gauss on large TFs and quantile=1.6, etc.) and decided to make a simple indicator:

GBPJPY 2018 pair.

Used:

1. CLOSE M1

2. sliding window - week (7200 values CLOSE M1)

3. simple MA

4. dispersion calculation according to the formula from this branch + quantile =1.6

Looking at:

A total of 7 trades would have been made in 2018 (+6/-1)

Total profit: +733 full points.

If anyone is sure this is the Grail, let them set up a TS and post it here. Right to the branch. Let the suffering people use.

What if you run it through the last 10 years? Is that possible? And find out where the series of profits in a linear regression converge to zero minus or plus?
 
Martin Cheguevara:
Well ... I hope the few people who know who I'm talking about and who really understand what I mean have understood. That's enough for me :)
If there will be decisions about adaptive asymmetry analysis, write a separate thread, as it is really just a fundamental thing, which will be effective not only in forex.
Who missed the discussion, see #6498 there everything you need to know about this branch. The question seems to be exhausted at this point in time.
I am ready to share with people who know my work to get results.
But this is unrealistic because the branches are quickly littered with grails of opinions, conjectures, forecasts and so on.

some do and some don't.

If you study, you won't want to write.

You will simply be misunderstood.

I am willing to share with people who know what works and that will give results.

;)

The highlighted context of your post is only wishful thinking, you will have to dig there yourself

 
Martin Cheguevara:
What if you run it for the last 10 years? Is that possible? And find out where the series of profits in a linear regression converges to zero minus or plus?

I don't have such archives... As a matter of principle I didn't use them, I thought it was all nonsense and worked with Erlang's tick flows... And lo and behold - how it all turns out...

 
Alexander_K2:

I listened and listened to CheGevara (about Gauss on large TFs and quantile=1.6, etc.) and decided to make a simple indicator:

GBPJPY 2018 pair.

Used:

1. CLOSE M1

2. sliding window - week (7200 values CLOSE M1)

3. simple MA

4. dispersion calculation according to the formula from this branch + quantile =1.6

Looking at:

A total of 7 trades would have been made in 2018 (+6/-1)

Total profit: +733 full points.

If anyone is sure this is the Grail, let them set up a TS and post it here. Right to the branch. Let the suffering people use.

I will send you an algorithm tonight which simply detects asymmetries, but there is a problem with the calculation period. What you have done is great, but it is unlikely to give the final result...because it is not really an asymmetry...