• From: Germany
  • Registered: 2012-02-17
  • Posts: 122

Topic: MonteCarloAnalysis

Hi,
very good Idear with the MonteCarloAnalysis. I think this is the right way. The biggest Problem in automatic strategy generation is the curvefitting.
We need more curvefitting tests.

But I think the monte carlo analysis is not enough. It is the right way but only the first step.

The product StrategyQuant have a monte carlo analysis too. But most of the strategies are curvefitted.
thomas

tnickel's Website

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

Re: MonteCarloAnalysis

tnickel wrote:

Hi,
very good Idear with the MonteCarloAnalysis. I think this is the right way. The biggest Problem in automatic strategy generation is the curvefitting.
We need more curvefitting tests.

But I think the monte carlo analysis is not enough. It is the right way but only the first step.

The product StrategyQuant have a monte carlo analysis too. But most of the strategies are curvefitted.
thomas

I suggest that you use the multi feature to establish robustness in addition to the MonteCarlo.

I posted a list of articles about robustness  http://forexsb.com/forum/topic/4961/robustness/

Blaiserboy's Website

  • Registered: 2013-05-27
  • Posts: 292

Re: MonteCarloAnalysis

I agree with tnickel. It's a step in the right direction but so far what we had in FSB was better.
My idea with the multi-market was different as well, I'll explain in another post.

  • Registered: 2013-05-27
  • Posts: 292

Re: MonteCarloAnalysis

I have the impression that monte carlo is changing parameters such as "shift", which apparently should not be changed by default as they have a enormous influence. Popov, can you please check and let us know which improvements are foreseen (in this forum some very nice ideas have been given).

Thanks
Nuno

  • Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 7,988

Re: MonteCarloAnalysis

I have the impression that monte carlo is changing parameters such as "shift",

Do you think so because some test lines starts later.

http://s29.postimg.org/ghaf9p1v7/screenshot_704.jpg

This is because of the "Randomize backtest starting bar". It skips a random number of bars from the data set.

The idea behind this test is to imitate different length of the data series. Also it can imitate different moments of beginning of the live trading.

Popov's Website

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

Re: MonteCarloAnalysis

This approach to Monte Carlo is superior to many that I have used because of the flexibility of the settings

Blaiserboy's Website