Warning nt7

Hello fellow traders . I was shocked when I had a loosing day with my automated strategy and after back testing for the same day NT7 showed a huge profit . So I did a Replay for the same day and the results were same as live trading -- loss no profit. I reloaded new NT7 and the results are the same. To eliminate that something wrong with my strategy codding I tested default strategy Sample MA crossover for March 27 CL futures and got - $-660 for back testing and $-20030 on the replay for the same time period . Its a huge difference and its makes any back testing on NT7 totally unreliable. Just pick a day and test any strategy as Back test and replay and you will see how huge are the difference .

Thats explains why so many strategies that shows great on back test are loosing in a real market.

I hope you can prove me wrong .

https://www.sendspace.com/file/qxc5cn

....

Its a huge difference and its makes any back testing on NT7 totally unreliable.

...

Thats explains why so many strategies that shows great on back test are loosing in a real market.

...

First of all you have to compare apples to apples (the same settings and reliable parameters, sometimes strategy adaptation to make reliable backtests, like adding an intrabar granularity for example). But, in general a long story short: it's almost impossible to achieve the same results in SA or even in the market replay mode as in the live trading. All you can do is to try to be as close to the reality as you can.

Strategy Analyzer can be a trap (and indeed, it is for many, as the history of posts showed). Deep understanding of SA is absolutely required before you can trust and rely on SA results...

Start from reading this:

http://www.ninjatrader.com/support/helpGuides/nt7/index.html?discrepancies_real_time_vs_bac.htm

and this:

http://www.investopedia.com/university/advanced-features-of-ninjatrader/strategy-analyzer.asp

and finally:

http://www.ninjatrader.com/support/forum/forumdisplay.php?f=17

try this:

https://www.sendspace.com/file/1psbiy

Put this cs in custom/type folder.

Then run the same strategy but in the strategy option you set "Tailor" in fill type (don't use default).

Have you in this way the same results as the replay mode?

Backtest is the waste of time, even historical will show you different results( actually worst than real trading). But I found market replay to give you 90% accurate picture.

Hello fellow traders . I was shocked when I had a loosing day with my automated strategy and after back testing for the same day NT7 showed a huge profit . So I did a Replay for the same day and the results were same as live trading -- loss no profit. I reloaded new NT7 and the results are the same. To eliminate that something wrong with my strategy codding I tested default strategy Sample MA crossover for March 27 CL futures and got - $-660 for back testing and $-20030 on the replay for the same time period . Its a huge difference and its makes any back testing on NT7 totally unreliable. Just pick a day and test any strategy as Back test and replay and you will see how huge are the difference .

Thats explains why so many strategies that shows great on back test are loosing in a real market.

I hope you can prove me wrong .

 

https://www.sendspace.com/file/qxc5cn

Thanks you all for reply. I wish NT7 instructions would be more clear about this problem. When I was dealing with MT4 the beck tests were more close to real .

Any way its great to have a forum like this to shear your ideas and frustrations.

I got reply from Ninja site ------ Thank you for your post.

This is to be expected when going from Real-time data to historical data. In Real-time we have the ability to run our strategies on a tick by tick basis. Back-test will only run once for each bar.

Thanks you all for reply. I wish NT7 instructions would be more clear about this problem. When I was dealing with MT4 the beck tests were more close to real .

Any way its great to have a forum like this to shear your ideas and frustrations.

 

I got reply from Ninja site ------ Thank you for your post.

 

This is to be expected when going from Real-time data to historical data. In Real-time we have the ability to run our strategies on a tick by tick basis. Back-test will only run once for each bar.

"Thank you for your post"

I'll translate. "We know we have a problem and are unwilling or unable to do anything about it. Neither are we willing to warn people of the problem. But rather let them find out the hard way, rather than risk losing any market share. We apologize for any inconvenience this may have caused."

W.

Free corporateese translations.