Trading advisers and robots - ***or is there somewhere magical?)) - page 8

 
Petros Shatakhtsyan:

Isn't it easy to write such an advisor? No~. It's an aerobatics job.

It's not top-notch, but it's not even worth the effort without logic.

a lot of people try to write something just for the fun of it. ....

and give them a contest in terms of - which system wins????

of course the winner is the one that at least in some way touches the subject of countering drawdown and at least slightly increases the probability of entering the market

 
danminin:

The main thing is to have an Expert Advisor, which once a month optimizes hundreds of your systems and selects the best one, and uses it to trade for a month.

And run this Expert Advisor in the tester for 5 years.

and then you will see whether such an approach is working or not.

Why does everyone focus on the duration of testering for years, I do not understand it at all, it even pisses me off, how many times I race, come to the drain - at some point it drains... I start from this point and it doesn't drain, it goes to a certain divergence again and then only drains, but it doesn't drain for a week or a day and if I reset it from the beginning of the day or week it may work out quite well... Although I didn't even manage to survive on automatic for a week...)))

 
Iaroslav Chornyi:

As a result of my personal practice and experience, I've learned for sure that there isn't a single advisor or indicator that can really help... it's even the opposite, it leads astray. As a rule, all of them are good only on history)))), but they are very good at filling beginners' heads with it)))) After all these attempts I've come to the conclusion that the market is almost like a living organism - no robot will tell you which hedge fund has decided to buy or sell assets. At the moment I'm testing based on Fibonacci .... there are no other indicators on charts, but my gut feeling is that I should add some simple technical analysis(trend lines, corridors, etc.) probably it will supplement the chart...))

Ha. I've been looking for the most reliable, like a dog, indicator since 2006. I've written a shitload of indicators.
Here is a screenshot of how Close[0] moves relative to Close[1].

Therefore, the end of almost any indicator may show both up and down direction. 50/50.
I have suffered through the suffering of indices that do not overdraw and do not interfere with each other.



 
Сергей Криушин:

Why does everyone focus on the duration of testering for years, I do not understand it at all, it even pisses me off, how many times I race, come to the drain - at some point it drains... I start from this point and it doesn't drain, it goes to a certain divergence again and then only drains, but it doesn't drain for a week or a day and if I reset it from the beginning of the day or week it may work out quite well... although i couldn't even make it through a week on automatic...)))

found

the usual sneak peek of history

 
Oleg Papkov:

Ha. I've been looking for the truest, dog-like indicator since 2006. I've written a lot of indices.
Here's a screenshot of how Close[0] moves relative to Close[1].

Therefore, the end of almost any indicator may show both up and down direction. 50/50.
I have suffered through the suffering of indices that do not overdraw and do not interfere with each other.



This is the truest prediction.... up or down will definitely go))))

 
Renat Akhtyamov:

found

normal history peeper...

No, it's honest - I'm surprised myself - absolutely stupid option, with your help by the way)))) endlessly opens and immediately closes, something like high-frequency trading, but by opening prices, it doesn't work on ticks... I can't understand anything myself... I don't have enough math or programming skills... I have it lying in the archive for a long time, I had the same story with it, now I added trawl from speed-1 hunter... but it doesn't work in the tester for real as well... doesn't open the right lines...

The picture

 
Vladimir Baskakov:
I've asked you a hundred times to do the monitoring, he'll throw you a lot of naysayers.

I'm not "denying it", I'm suggesting you do it yourself! The regcodes - no problem.

Come on!

 
danminin:

Or write an EA that once a month optimises hundreds of your systems and chooses the best one out of them, and trades on it for a month.

And run this EA in the tester for 5 years.

and then you will see whether this approach works or not.

How on earth can you write an EA if - I'll tell you again - the choice is intuitive ?!!

But the fact that I haven't lost anything for five years (and consequently CA earns money on me all the time) - says that the method is quite realistic. The only thing I need is to convert it from intuitive to formalized. Unfortunately, I still have problems with this.

 
Petros Shatakhtsyan:

Isn't it easy to write such an advisor? No~. It's a matter of aerobatics.

The main thing is not to write, but to formulate the rules of selection. I can't do shit...

 
Сергей Криушин:

Why does everyone focus on the duration of testering for years, I do not understand it at all, it even pisses me off, how many times I race, come to the drain - at some point it drains... I start from this point and it doesn't drain, it goes to a certain divergence again and then only drains, but it doesn't drain for a week or a day and if I reset it from the beginning of the day or week it may work out quite well... Although I couldn't even make it through a week on automatic...))

That's what I'm talking about !

I've said it more than once - ANY system has periods of loss that are longer than periods of gain, which means that any system is losing in the long run. And the key to profit is just to constantly change the systems that don't work for those that still do.