1 Topic by mkowa (edited by mkowa 2018-05-20 13:43:07)

  • mkowa
  • Licensed Member
  • Offline
  • From: Germany
  • Registered: 2018-05-20
  • Posts: 8

Topic: Any success stories?

Hi I am new to this forum, bought the use of this software and the last week I did nothing else than trying. My family does not see me very often now.

But everything I found is that obviously every strategy fails the time after the backtesting. Seems all of them are curve fitted.
It is hard to believe that a strategy wich would have worked over years stops working the moment after the testing period! But that's what it looks like! Monte Carlo does not much to that either!

As I am still testing this piece of fine work (and am still able to decide about quitting) I would like to hear if there is someone, who found profitable strategies this way and is actually making money with them.

If so: What is the secret of finding those strategies?

Please convince me to stay! I admire this great software!

  • Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 8,013

Re: Any success stories?

What is the secret of finding those strategies?

Store all strategies you have generated and tested. After several weeks re-test the strategies with new market data. You will find that 5% - 30% (depending on your experience) of these strategies make money on the unseen data. Choose the best of them and run on a demo or on a real account with minimum investment. You can slowly scale the entry amount on the winning strategies. By meantime continue generating and testing new strategies. You will always have fresh replacements in that way.

The forex trading is a serious job and needs commitment to get it right. Please note that you will make money only when you outperform the other traders on the market. It is only possible with experience, patience and the best tools for the job.

Popov's Website

3 Reply by sleytus (edited by sleytus 2018-05-20 21:23:45)

  • Registered: 2017-03-14
  • Posts: 709

Re: Any success stories?

Popov wrote:

You will find that 5% - 30% (depending on your experience) of these strategies make money on the unseen data.

Why do you think 70-95% of the generated strategies that make it into a collection (and whose stats are probably very good) do not perform well?  Is it because there is something defective with the strategy's logic or settings (in which case that strategy should be permanently deleted)?  Or do you think the strategy could perform better in the future when the data pattern changes (in which case it is worth saving the strategy and revisiting it at a later time)?  In other words -- if a strategy does not currently perform well then do you still keep it or do you delete it?

I understand this is a difficult question to answer and there may not be a definitive answer -- I'm just curious as to what your impression is. 

My experience suggests the answer is probably closer to the 2nd explanation -- strategies perform like a roller coaster.  Using large data horizons doesn't help since that mostly skews optimization toward old, stale patterns and hides the details of what is happening now.  I mean, when you use large data horizons the strategy may currently be performing like a roller coaster but you can't see it because the last few months of data are obscured by years of historical data.

Thanks

  • Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 8,013

Re: Any success stories?

The biggest problem is the curve-fitting. The Generator and Optimizer are so powerful that tend to exploit every possible market movement of the provided testing data. Monte Carlo and Walk Forward can filter the most obviously over-optimized strategies, but the rest can be only detected on unseen data.

Your method for composing history data from various symbols may improve the initial robustness of the collection.

Popov's Website

  • Registered: 2017-06-07
  • Posts: 128

Re: Any success stories?

how did you do retest. when i started reactor to find strategies from 2010-2015 and i get some strategies. now i want to test these strategies with data from 2010-2018 . did i have to put the collection in the validator or how does it work

  • Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 8,013

Re: Any success stories?

There are two ways:
- Load the Collection and "Recalcualte" with the new data.
- Use the Validator to calculate + test and validate.

Popov's Website

  • Registered: 2017-06-07
  • Posts: 128

Re: Any success stories?

Popov wrote:

There are two ways:
- Load the Collection and "Recalcualte" with the new data.
- Use the Validator to calculate + test and validate.

When i use recalculate than eastudio calculate with new data and try to find best results with changing indicator parameters or would all be same

8 Reply by rjectweb (edited by rjectweb 2018-05-22 20:34:08)

  • Registered: 2017-05-06
  • Posts: 51

Re: Any success stories?

Hi,

just a quick post to show it is completely possible to build good strategies with Popov's tools.

This is one of my first demo accounts with a Portfolio EA built with EA Studio. This account is going to be a year of demo quite soon. Low leverage, around 9% profit and 2.5% DD opening 0.01 positions (would have been 90%profit to 25%DD in a year).

This portfolio was one of my firsts. No special conditions/filters were applied to build the portfolio (most of the strategies directly out of the box) and the POrtfolio EA hasn't been pruned until quite recently when I removed some of my worst strategies (the results could have been better if I had pruned the portfolio frequently). Of course I have other accounts where my results were bad, but this one proves it is very possible to build good and durable strategies.

Just showing this results here to encourage others and showing EA Studio is a great tool.

https://s9.postimg.cc/n0y3v1zez/Mix02_3.png

Warmest regards from Spain
RJ

9 Reply by moreirajbn (edited by moreirajbn 2018-05-23 19:42:57)

  • Registered: 2018-05-01
  • Posts: 5

Re: Any success stories?

Popov wrote:

The biggest problem is the curve-fitting. The Generator and Optimizer are so powerful that tend to exploit every possible market movement of the provided testing data. Monte Carlo and Walk Forward can filter the most obviously over-optimized strategies, but the rest can be only detected on unseen data.

Your method for composing history data from various symbols may improve the initial robustness of the collection.




Hi, are you able to tell when walk forward analysis will be available? I mean, true walk forward with multiple rolling periods....

Thank you

  • Registered: 2020-10-15
  • Posts: 4

Re: Any success stories?

I've been practicing for about 6 months. I still have both wins and losses. I'm curious to hear other people's success stories.

  • Registered: 2020-04-12
  • Posts: 7

Re: Any success stories?

jandkaymusic wrote:

I've been practicing for about 6 months. I still have both wins and losses. I'm curious to hear other people's success stories.

Statistics of the brokes say that 80-90% of traders lose money.
I haven't read success stories here and I don't even know if anyone uses it on real accounts.