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

Topic: Question about using Multi-Market when backtesting...

Suppose I'm generating EURUSD strategies and using a DataHorizon of 1 year.

And then I include Multi-Market testing and require that my strategies pass 3 additional symbols -- e.g. USDCHF, GBPUSD and USDJPY.

My question is then -- how is that different than using a DataHorizon of 4 years and not including Multi-Market testing?  I mean, when you include Multi-Market testing isn't it roughly equivalent to increasing the DataHorizon?  Or are the signal patterns from different symbols significantly different that Multi-Market testing stress-tests your strategies in ways that increasing the DataHorizon can not?

Thanks for any insight you may have...

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

Re: Question about using Multi-Market when backtesting...

Each pair has unique patterns.

Blaiserboy's Website

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

Re: Question about using Multi-Market when backtesting...

Blaiserboy wrote:

Each pair has unique patterns.

Just to be clear -- are you saying those patterns (say USDCHF) would not appear in EURUSD even if I increased the DataHorizon of my EURUSD backtesting?   I assume you'll say -- yes, that's correct.

But then my next question is -- if there are USDCHF patterns that don't occur in EURUSD then why would I want to test my EURUSD strategies against USDCHF data in the first place?

I understand the goal is robustness and I don't have anything personal against Multi-Market, it's just that when including Multi-Market it feels like I end up throwing away many strategies I bet would have traded just fine in a live account.

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

Re: Question about using Multi-Market when backtesting...

The price bars are influenced by cycles...... many kinds of cycles, economic, time, etc

Yes you can find same patterns on many pairs, however, they will appear in different sequences, each pair has iots own market ie investors, fundamentals, reaction to other pairs.....

Correlation can appear for brief times and then it is no more.... perhaps you would overlay a few pairs on a chart to watcvh what occurs.

Blaiserboy's Website

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

Re: Question about using Multi-Market when backtesting...

Okay -- I think I get it.  This helps a lot -- thank you.