Pair trading and multicurrency arbitrage. The showdown. - page 285

New comment
 
Irina Kolosova #:
There is a salt. You just can't just take two indices and work on them, you need to take 5 different instruments and trade against one. Just two instruments do not work accurately, today Dax is faster, tomorrow it will be faster than Nazdak....

to calculate it, there's 100 securities, I think.

 
Renat Akhtyamov #:

to count it, there's 100 papers, I think.

Well, we don't only look at indices( There is gold, for example, as a filter...).
 
Irina Kolosova #:
Well, we don't only look at indices( There is gold, for example, as a filter...).
You probably don't need an index for such things. I can't even imagine where you can plug it in.
A cluster is made of instruments that are traded.
The optimal variant of a cluster is currency bets, and all the crosses between them to fix the overflow.
and the filter there is as simple as a rifle AR15^
the simplest filter, but at the same time the most effective - is what happens to the cluster: the volume of the whole cluster grows or this volume shrinks.
If you find the answer - the cluster volume grows or shrinks, then part of the problem is already solved, you have an understanding of what to open and what to close at this moment - whether it is buying or selling.
There is a signal to open and close positions in opposite directions depending on the found answer in the filter problem.
 
Vladislav Vidiukov #:
This indicator shows a divergence, but it does not mean that there will be a convergence afterwards
The assumption that an indicator showing divergence will eventually lead to a convergence is exactly how retail accounts get obliterated trying to trade synthetic pairs with amateur algorithms, true multicurrency arbitrage and pair trading cannot be mathematically validated by simply throwing a basic overlay indicator on a chart and blindly hoping the lines cross back, because standard MetaTrader indicators do not calculate the actual statistical cointegration of the assets. If your architecture is not running a continuous Augmented Dickey-Fuller test in the background to mathematically prove the stationarity of the spread before executing the entry, you are not actually doing pair trading, you are just gambling on correlated noise and letting the broker's hidden synthetic slippage slowly eat your entire margin
 
Amanda Vitoria De Paula Pereira #:
The assumption that an indicator showing divergence will eventually lead to convergence is exactly how retail accounts are destroyed when attempting to trade synthetic pairs using amateur algorithms. True multi-currency arbitrage and pair trading cannot be mathematically justified by simply overlaying a basic indicator on a chart and blindly hoping that the lines will cross, because standard MetaTrader indicators do not calculate real statistical asset cointegration. If your architecture does not perform a continuous Augmented Dickey-Fuller test in the background to mathematically prove the stationarity of the spread before executing an entry, you are not really pair trading, you are just playing on correlated noise and letting the broker's hidden synthetic slippage slowly eat up all your margin.
even with this test, pairs trading doesn't work very well
 
Irina Kolosova #:

I want to apologise. We raised everything, the account on the last screenshot is FC Real, there the execution was about 370ms (one way), I will not say more precisely, the account was deleted, now a new account. But, we launched two more accounts in parallel, for testing (we copied signals, like news signals, who remembers), on Alpari and Alfa.... If you're interested, let's look for Alpari. Alfa is a demo, but the demo on AlfaForex corresponds to the real one in terms of execution. We were surprised that with Alfa's horse spreads and 1000ms execution (one way) all trades (except one) closed in plus. The spread on Alfa is simply murderous.


Irina, try to test it not only on nonnicks.

The result is interesting.

I'm sure that some period will have to be excluded, but it's not about the noughts, as I think.

 

here


 
Renat Akhtyamov #:

Irina, try testing on more than just nones.

It's interesting to see the results.

I'm sure that some period will have to be excluded, but it's not about the nones, as I think.

Greetings! On the nonfarm, big volumes and huge momentum in different directions. Also on the release of US inflation data.
 
Irina Kolosova #:
Greetings! On Nonfarms, big volumes and huge momentum in different directions. Also on the release of US inflation data.

Match it to the day of the week.

Which one is it?

https://www.mql5.com/ru/forum/448777/page278#comment_59423713

 
Renat Akhtyamov #:

match the day of the week

which one is it?

https://www.mql5.com/ru/forum/448777/page278#comment_59423713

It was the nonkeys.
1...278279280281282283284285
New comment