1 Topic by Irmantas (edited by Irmantas 2017-10-05 15:40:46)

  • Registered: 2016-05-16
  • Posts: 190

Topic: Enter Once Strategies too optimistic results

Hi,
I generated very good looking strategy, however I think it is too good to be true smile There is nothing magic about "enter once per week" on the round number. Similar results on EUR USD too. Changing that "one per week" good results disappear. Also mt4 tester do not show similar backtest. Can someone please check these indicators for mistakes? I believe there should be one. Or there is no mistakes and it is real holy grail? smile THANKS!

Post's attachments

200015 EURJPY 1440.xml 11.62 kb, 5 downloads since 2017-10-05 

You don't have the permssions to download the attachments of this post.
  • Registered: 2016-05-16
  • Posts: 190

Re: Enter Once Strategies too optimistic results

Well it seems changing Method from "Pessimistic" to "Optimistic" destroys profits. hmm, should it not be opposite way? Well, I guess there is no mistakes in indicators, just ambiguous bars playing out strangely.

  • Registered: 2016-05-16
  • Posts: 190

Re: Enter Once Strategies too optimistic results

Even more strangely, with different data, good results remain with all interpolation methods... big_smile So there is some big confusion. On my broker data, good results is destroyed when changing from "pessimistic" to "optimistic", but with Ducascopy data it remains all good. Where is a problem?

https://s1.postimg.org/1d1wbiegob/problem.png

  • Blaiserboy
  • Blaiserboy
  • Junior Part-Time Aspiring Space Cadet
  • Offline
  • From: Toronto
  • Registered: 2010-09-06
  • Posts: 2,849

Re: Enter Once Strategies too optimistic results

you can use filters to ensure you have good stuff.... and you can also use walk forward

Blaiserboy's Website

5 Reply by Irmantas (edited by Irmantas 2017-10-05 16:06:12)

  • Registered: 2016-05-16
  • Posts: 190

Re: Enter Once Strategies too optimistic results

Sorry Blaiserboy, it seems you do not read what I write here. No walk forward will solve bogus indicator or some strange mesh up with different data or interpolation methods. It is very dangerous to get too optimistic results and to clear things out is a must.

  • Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 8,092

Re: Enter Once Strategies too optimistic results

It happens that the strategy may take a great advantage of exploiting the interpolation algorithm. I call that algorithm-fitting. This problem appeared in the first year of creating and testing FSB. The only solution I found was to add different algorithms that interpolate the bars in different way.

The Comparator tool calculates a strategy by using all selected algorithms and plots the result on a chart. 

https://s1.postimg.org/1hcazdoo8v/screenshot_9.png

We see that this strategy is a perfect example of exploitation of the Pessimistic algorithm. We see it shows the best result. Shame for it smile.

Fortunately the backtest shows ambiguous bars, which is a clear sign that there is something wrong. However it is possible to have a case for algorithm-fitting without ambiguous bars.

It is always a good idea to use the Comparator.

In that connection imagine how the MetaTrader users detect such cases.

Popov's Website

7 Reply by Irmantas (edited by Irmantas 2017-10-05 16:20:34)

  • Registered: 2016-05-16
  • Posts: 190

Re: Enter Once Strategies too optimistic results

Thanks for answer. Firstly I used DucasCopy data, and Comparator test showed nothing bad... Lesson: always test on few different data sets, and look in Comparator smile But firstly, always get suspicious when you see very good results.  Have a nice day!