1 Topic by infernal199 (edited by Popov 2015-09-20 06:08:48)

  • Registered: 2015-09-20
  • Posts: 4

Topic: Why bother creating strategy for the previous month?

It sound strange but here is the prob....

I used FSB Demo data Full 2014 for H4 chart to build a strategy & it created a awesome strategy with 3.60+ System quality & also 400% ROE gain for the whole year!!

Okzz

Now I tested same strategy on 2015 Jan to April but now It showing within first 3 months It blow up the whole $1000 account!!

Note: I used FSB for backtest, its super fast though!!

My ques is.... If strategy cant even prove it self for next 1month at least then why we will be even bother to create a strategy that can gain thousands of $$$ in previous month??

  • Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 8,024

Re: Why bother creating strategy for the previous month?

It is called OOS - Out Of Sample testing. Some traders think that if a strategy works well on a past period and it confirms the performance on the new data, it proves that it is not over-optimized.

However, the OOS doesn't prove that the strategy is not over-optimized for the full period.

See this topic for more info: OOS question

Popov's Website

  • Registered: 2015-09-20
  • Posts: 4

Re: Why bother creating strategy for the previous month?

thx again.. we are now discussing in both forum

Ok... If I use OOS 20% or 40% will it be reasonable for better strategy?

Lets say for M30 chart I use 3000 data bars with 40% OOS

& If i want to use this strategy for next week from Sep 21 to 25 will it be reasonable?

  • Registered: 2016-02-14
  • Posts: 160

Re: Why bother creating strategy for the previous month?

infernal199 wrote:

It sound strange but here is the prob....

I used FSB Demo data Full 2014 for H4 chart to build a strategy & it created a awesome strategy with 3.60+ System quality & also 400% ROE gain for the whole year!!

Okzz

Now I tested same strategy on 2015 Jan to April but now It showing within first 3 months It blow up the whole $1000 account!!

Note: I used FSB for backtest, its super fast though!!

My ques is.... If strategy cant even prove it self for next 1month at least then why we will be even bother to create a strategy that can gain thousands of $$$ in previous month??

strategy building is process and there are many steps.

if the strategy loses 1000s usd in the previous month then you can forget about the strategy.

if it makes 1000s in the previous testing period then you test it on some other period.

as you said it is superfast-everything in FSB is super fast: you can change a indicator or change the parameters of and indicators and it will show the result on your balance immediately without pressing a button or even changing the page.

FSb is not perfect...nothing is....but it gives you results in a blink of a eye..and in real time