A topic for traders. - page 35

 
Georgiy Merts #:

Oh... And someone else is making fun of my League... I've got five times fewer robots... Really, my trading computer is 2001, 1Gig of memory. Four terminals...

I wonder how those 5 thousand robots deal with 500 positions limit?

Nobody is laughing at your league anymore, they're crying.
 
Vladimir Baskakov #:
Your league hasn't been laughed at for a long time, it's been cried at.

Crying??? I hadn't noticed. Well, okay, crying. So I've only got a thousand robots, and the man over there has five thousand !

 
Georgiy Merts #:

Crying??? I hadn't noticed. Well, okay, crying. So I've only got a thousand robots, and the man over there has five thousand !

Form a party.
 
Vladimir Baskakov #:
Create a party.

Yuck... And why would the owner of 5,000 robots need some poor relative with only one? You, Vladimir, are always talking nonsense.

 
Georgiy Merts #:

Oh... And someone else is making fun of my League... I've got five times fewer robots... Really, my trading computer is 2001, 1Gig of memory. Four terminals...

I wonder how those 5 thousand robots deal with 500 positions maximum.

8 basic algorithms, up to 1000 variants of each (differences not only in parameters), 4 symbols (FORTS).

I5-2500K, 32GB - 3 cores and 6 GB allocated for trading virtual machine.

Test trade - trades are not executed, but counted as if they were executed. Not "history testing" - because the information is used, which is not stored in the history. Only the basic algorithms are tested on the history (over 3-4 years), before being admitted to the test trading.

Those, who have proven themselves successfully in the test (at least 3 years) - are accepted for real trading. For new modifications of the same basic algorithm (usually a hybrid of the most successful variants) the period to the real account is shorter - from one year.

At the moment there are 17 robots on the real market - 4 basic algorithms, 3 symbols.

Best result in the test so far is about 100% of RO per year on average over 3 years (fixed lot = 1 contract), with maximal drawdown around 30% of RO.

In real trading, trading one or two RTS contracts, he has earned more than 130$ for almost 2 years, with max drawdown less than 20$.

 
JRandomTrader #:

8 basic algorithms, up to 1000 variants of each (differences not only in parameters), 4 symbols (FORTS).

I5-2500K, 32GB - 3 cores and 6 GB given to trading virtual.

Test trade - trades are not executed, but counted as if they were executed. Not "history testing" - because the information is used, which is not stored in the history. Only the basic algorithms are tested on the history (over 3-4 years), before being admitted to the test trading.

Those, who have proven themselves successfully in the test (at least 3 years) - are accepted for real trading. For new modifications of the same basic algorithm (usually a hybrid of the most successful variants) the period to the real account is shorter - from one year.

At the moment there are 17 robots on the real market - 4 basic algorithms, 3 symbols.

Best result in the test so far is about 100% of RO per year on average over 3 years (fixed lot = 1 contract), with maximal drawdown around 30% of RO.

However, in real trade, trading one or two RTS contracts, he has earned more than 130$ for almost 2 years, with max drawdown less than 20$.

I see. Well, if trades are not opened, but only counted, then of course there will be no conflict.

Yes, everything is essentially the same as in my League. The idea is the same...

 
Georgiy Merts #:

Yes, it's essentially the same as mine in the League. The idea is the same...

Not exactly. I don't envisage a permanent rotation. I have my sights set on finding options that can work profitably for at least a few years.

By the way, that (so far) best option shows good (though worse) results on both Si and MIX, and stands them on real trade too.

 
JRandomTrader #:

Not really. I don't envisage a permanent rotation. I have my sights set on finding options that can work profitably for at least a few years.

Well, well... That's where I started, too... Unfortunately, I see that no TC lasts longer than half a year. And it's wise to switch them in advance.

However, I have only forex pairs. On other symbols, perhaps the situation will be different.

 
One of my variants got into test with RTS-12.16 and into real with RTS-12.19 and is still trading successfully on the real.
 

Existing profitable TCs can surely be improved by looking at them with an understanding of new changes in the nature of markets.

The nature of markets is constantly changing from the international environment, rulers of states leading countries to decline or prosperity.

There are patterns in TA, but they do not always take into account all the nuances within a certain time frame. This is why some groups of TA may not take into account many external stimuli and go with the old programme.