What do you think about a backtesting with 99% of model quality data? - page 2

New comment
 
bearnaked:


Raptor, have you sure?

 I am asking that, because looks like that the tool that I use, allow simulate slippage and also spread variable. I won't say the name of the software, but is a famous software around internet.

I am as sure as I can be,  but if you want to make sure for yourself ask Birt.  Spread can be variable because the fxt file holds Bid and Ask prices and the tick data also hold Bid and Ask prices,  so when the fxt file is made there is the option to incorporate the spread from the tick data . . .  how is slippage facilitated ?
 

ubzen:
Ok, so which year did you perform the optimization on [if any]?
I didn't do optimization yet.. but the backtesting is from 2011 till 2012/(may). I Also tested in 2012/2013 and the result is also good.

 

But the question is, is possible to trust in 99% model data? 

 

But the question is, is possible to trust in 99% model data? 

What do you mean by trust?

Model Quality does not equal testing Quality. Example I can test with 90% Modeling but trade Once_Per_Bar on Open_Only.

Model Quality does not mean good data. Example the model could be 99% however the data could be Fake.

Model Quality does not equal Future Results. System could work on 999999% quality but still fail in live markets.

So what exactly are you asking me for.

Do I trust myself using 99% modeling quality, because I know what I'm doing?

Do I trust someone else results because they're using 99% model?

What do you mean by trust? 

 
ubzen:

But the question is, is possible to trust in 99% model data? 

What do you mean by trust?

Model Quality does not equal testing Quality. Example I can test with 90% Modeling but trade Once_Per_Bar on Open_Only.

Model Quality does not mean good data. Example the model could be 99% however the data could be Fake.

Model Quality does not equal Future Results. System could work on 999999% quality but still fail in live markets.

So what exactly are you asking me for.

Do I trust myself using 99% modeling quality, because I know what I'm doing?

Do I trust someone else results because they're using 99% model?

What do you mean by trust? 



Ubzen.

 

Well.. I  was asking, if is possible to trust that we can get similar result in live testing... comparated with backtesting with 99% model.

 

I want to hear your experience guys...

 

for example.. "well.. in backtesting I get an excelent result with 99% of model.. but in live testing (real/demo) the result is totally different.. bla bla bla. 

 
RaptorUK:
I am as sure as I can be,  but if you want to make sure for yourself ask Birt.  Spread can be variable because the fxt file holds Bid and Ask prices and the tick data also hold Bid and Ask prices,  so when the fxt file is made there is the option to incorporate the spread from the tick data . . .  how is slippage facilitated ?


ok. thank you Raptor
 
bearnaked: Ubzen. Well.. I  was asking, if is possible to trust that we can get similar result in live testing... comparated with backtesting with 99% model.I want to hear your experience guys...for example.. "well.. in backtesting I get an excelent result with 99% of model.. but in live testing (real/demo) the result is totally different.. bla bla bla. 
I don't think there's any direct correlation between modeling quality and future results. This is why I recommend doing a Live test on my first post.
 
ubzen:
I don't think there's any direct correlation between modeling quality and future results. This is why I recommend doing a Live test on my first post.


Ok.. thanks bro

 

I just put it live... the result is good.. but.. just started... no data enough.

 

Thank you all guys.. and good luck  

 

 

Hi eriaku, I am trying to backtest the strategy of gold through MT4. But i cannot use strategy optimization option. when i click, nothing happens, please please help me how to back test strategy.
 
bearnaked:


Ok.. thanks bro

 I just put it live... the result is good.. but.. just started... no data enough.

Your SL & TP give you a Risk:Reward of 30:5 meaning you need a BE Win Rate of 30/35 = 86%,  looking at your Average loss vs Average win you have 28.97:6.01 this gives a BE WR of 28.97/34.98  =  83%.  Your actual WR was 85.14%  in my opinion this is too close to what you need to break even,  if the spread is a little worse live compared to your Strategy Test or you experience some slippage you will fall below your BE WR . . .
 
What I have learned form the testing is: It does not matter if it is a backtesting or forward testing.  The market is always unique.  What I mean by that is the winning path you obtained is unique too.  A minor setting change in your test can lead to a totally different results.  The market always have the upper hand.  It waits you to show the card and then he shows his.  At the best, you only stands 1/2 chance to win (not to count spread and rollover etc).  That is the whole problem.  The best predictive property of the market as I see it, is the major trend, if it exists,  espeically on the daily chart and above, do not trade against it.  The major trend is not easy to change, it reflects the overall economical balance between the two countries.  that is why, the break of a trendline is a significant event.  I think only the EAs uses trendlines stands a better chance to win.
123
New comment
Reason: