Backtester Bug?

[mailro]

Hi,
I have a trading system that was originaly designed in Amibroker and it was giving an 80% success. At ninja backtester it gives 100%.
It enters and exits the trades intrabar. Is that possible to conflict the ninja backtester?
I am sending you a photo.

[NinjaTrader_RyanM]

Hello mailro,

Thank you for your post.

You can get different results when comparing backtesting systems. This could be due to different data, fill algorithms, or framework.


How have you structered the conversion from Amibroker to NinjaScript?

[mailro]

What do you mean how I structured it? I just wrote it in nijascript.
If I got it right there is no bug in ninja backtester the results are trusty.
The data actually differ because amibroker gets data from strategy runner since ninja doesn't have DDE

[NinjaTrader_RyanM]

My first reply was meant to indicate some general reasons for seeing discrepencies across different systems. We need to know how your code is structured to help identify a specific reason why you're seeing that 100% win rate. How are your exits setup? Do you have conditions and Exit statements or are you using only SetProfitTarget() for your exits?

[mailro]

The code goes like this:
Once the price cross (something) enter a position short or long.
Once you enter a position set a profit target and a stop loss target

The code has been created using the strategy wizard.
That's it.
**this (something) doesn't have any optimization values and didn't went through any optimization to get created.

Anyway I will let it ride for a couple of days untill some losses appear and I will let you know so you can locate the bug to the backtester.

[NinjaTrader_RyanM]

Hello Mailro,

We never claim that backtesting results will match performance in a live trading environment.

Please see this article on discrepencies between real time and a backtest.

[mailro]

Of course I know and I understand this.
If I see anythig strange between the backtester trades and the actuall trades I will let you know.

Thanks again for your help

[mailro]

Dear Ryan there several bugs with the backtester that why I was getting the 100% results.
I used as a fill algorithm the default.
The bugs are:
1) The backtester considers some trades that don't occur in Real life
2) Some Real life trades occur at different prices than the backtester considers
3) Some Real life losing trades are been reported as winning ones at the backtester
4) Some Real life losing trades are not been reported at all

If you are planning to resolve these bugs on a future version let me know if you want me to send you the trade lists or the charts of backtester and sim accounts so you locate the bugs more easily.

[NinjaTrader_RyanM]

Hello Mailro,

We expect there to be differences between real life trading and backtesting results. Some of the reasons for this are covered in the link in my previous response. Additionally you could have different trade groupings depending on when you start the test.

Can you provide a scenario so we can attempt to reproduce your findings?

[mailro]

ok,
I have send the scenario at support@ninjatrader.com