Inconsistent Backtesting

[s.jussen]

Hi;

When a colleague and I have independently backtested the same strategy, NT spits out significantly different results. On one occasion, we both downloaded the data from FXCM and on another, we both imported third party data. Each test produced varying results. Furthermore, an analysis of the data from FXCM showed an average difference of 1pip for each row of data.

1. If we are both using the same strategy, parameters and data, why are we receiving different strategy results?
2. If we both import the same data for the same time frame from the same source (FXCM), why is the date in the files themselves different from one another?
3. If this is the case throughout, how can any backtesting in NT be considered reliable?

Thanks to all in advance.

[NinjaTrader_ChelseaB]

Hello s.jussen,

A good place to start would be printing the bar time (and optionally the Open, High, Low, and Close) to see that the data seen by both strategies is exactly the same.

For example:

Print(Time[0]+" | "+Open[0]+" - "+High[0]+" - "+Low[0]+" - "+Close[0]);

Prints will go to the output window. (Tools -> Output Window...)

Can you save these from each computer? (Right-click output window -> select Save As... -> give unique names -> click Save)

Once saved, please attach both files to your next post.

[s.jussen]

Print(Time[0]+" | "+Open[0]+" - "+High[0]+" - "+Low[0]+" - "+Close[0]);

You want me to add this into the code itself? Or is there an easier way to execute this demand?

[NinjaTrader_ChelseaB]

Hello s.jussen,

The idea is to add the print to the strategy code so that you can understand why the strategy code is behaving differently between machines.

The first step of which is ensuring the data is exactly the same between the machines by printing the bars the strategy is processing.

Please add these prints, save the output, and then post these in your next post.