A topic for traders. - page 151

 
Aleksei Stepanenko #:

I wonder how?


What do you mean?

On each row, you build points on the zigzag (and rather not at random, but you have to pre-select the step of volatility for each pair. For example EURUSD = 100 5-digit pips, while GBPUSD zigzag step will dynamically change from EURGBP.... somehow ... I don't know ... probably nonsense ... but there is a kind of feeling, that pairs should be somehow 'leveled' beforehand )

And then we should normalize the series (or remove the trend from the series, the same thing) but take not all points in a row, but only those extremums which coincide in time.


 
Evgeniy Chumakov #:


What do you mean?

On each row, you build points on the zigzag (and rather than on a whim, you have to pre-select the volatility step for each pair. For example EURUSD step = 100 5 digits, GBPUSD zigzag step will dynamically change from EURGBP.... somehow ... I don't know ... probably nonsense ... but there is a kind of feeling, that pairs should be somehow 'leveled' beforehand )

And then we normalize the series (or remove the trend from the series, the same thing) but take not all points in a row, but only those extremums which coincide in time.


How do you level AudChf(0.657) and. NzdJpy(77.00)?
 

And this is what the zigzag looks like after removing the trend. Volodya's favourite waves.

 
Aleksei Stepanenko #:

Vitaly, what is the theoretical premise for a synthetic? Do they have to converge or what? I'm not up to speed on this subject, but for them to converge, you have to claim that there is a relationship between them, i.e. you have to see it with your eyes.

?

Many pairs are interrelated, but they converge in small pips and diverge in large pips. You can open orders and it is not certain that prices will converge. Of course, they can converge and oscillate around the average but surely the moment will come when they will diverge greatly and the price will fall much in the red. One can trade by correlation, but in the long term it means loss of the deposit. For example, EURUSD and GBPUSD have diverged too much.
 
Vladimir Baskakov #:
How do you level AudChf(0.657) and. NzdJpy(77.00)?


I don't know ) I haven't written anything about these pairs ) Suggest your own option.

 
Evgeniy Chumakov #:

And this is what the zigzag looks like after removing the trend. Volodya's favourite waves.

it's not so much that you've removed the trend as that you've integrated the noise

 
Maxim Kuznetsov #:

it's not so much that you've removed the trend as that you've integrated the noise


What is integrating noise?

 

To compare AudChf and NzdJpy we should look for something in common between them in the model we will use ?

They either have USD or deposit currency in common, and usually they are the same. In other words, in order to "compare" you should add operations with majors, because it is better to know the point price and calculate profit/loss of one object.
Then we can carefully build both charts in common coordinates.

for example, take April 1 as a starting point, assume that you bought both pairs at once for equal amounts, and then look at the daily and hourly result per currency of deposit (or usd).

X represents time on the X axis and Y represents profit/loss in usd. Both graphs are now comparable values

 
Evgeniy Chumakov #:


What is a noise reintegration?

folded repeatedly.

 
If the correlation coefficient is 0.9, it means that pairs can, for example, converge/diverge in 90% of cases by 0.001 points and in 10% they can diverge by a great distance. As the result, 90% of 100% reversals will be 0.1 point and 10% will cause a 100 point loss. It seems silly to trade by correlation expecting to gain 1 pip and risk losing 100
Reason: