Backtesting strategies on ES ##-##

[TraderLeo]

I admit I am pretty new at backtesting but I had no difficulties on developing a couple of strategies and backtesting them on ES 03-17.
Now I'd like to backtest them on ES ##-## for the last 12-24 months. I tried everything but to no avail.
I have an active trading account (not a demo..) and Continuum Data feed.
Can you provide me with a clear step by step procedure in order to backtest my strategies on that instrument?
Thank You very much.
Leo

[NinjaTrader_Jesse]

Hello,

Thank you for the post.

Testing this symbol would use the same set of steps in the Strategy Analyzer as any other instrument but this does depend on the data provider.

NinjaTrader brokerage doesn't provide historical or live data for the continuous contract symbol, but because ninjaTrader comes with the "auto merge" option enabled, selecting the current contract and using MergeBackAdjusted would provide the same information you would get from the continuous contract. Other providers like Kinetick do offer the ##-## symbol and in that case the symbol could be used in the analyzer.

I look forward to being of further assistance.

[TraderLeo]

Thank You Jesse,
it works !
I am backtesting on ES 03-17 back to all 2016 and beyond.

Just a further question: is there any difference on how data are computed between the present contract data ( from Jan 2017 to present ) and the older data ( ie. from Jan 2016 to Dec 2016) ?

For your info I am backtesting a strategy based on 3-min data chart.

Thank you very much.

Leo

[NinjaTrader_Jesse]

Hello TraderLeo,

I am not certain I understand the question, are you asking if the methods of recording the data have changed or are there differences between 7 and 8? could you provide more detail on what you are seeing?


I look forward to being of further assistance.

[TraderLeo]

Thank You for your quick reply !
I'd like to be sure that data provided by Continuum to back-test ES 17-03 for previous periods ( ie. before the roll over date..) "have the same format and behavior" than the data of the present period from 12/08/16 to 2/21/17.
I am asking this probably silly question because I am finding a VERY VERY different behavior of my strategy when back-testing on actual data than on merged data.
I will of course deeply review my strategy but I'd like also to be sure that back-testing on not merged data is same than back-testing on merged data.
Hope I have explained myself correctly.
Thanks for your help.
Leo

[NinjaTrader_Jesse]

Hello,

Thank you for the reply.

To further clarify, you said, "I am finding a VERY VERY different behavior of my strategy when back-testing on actual data than on merged data."

Both of these cases would use "actual" data, do you mean that one test using Merged data is different than a test using Non-merged data? If so I would expect this. Could you confirm if this is the case with this question?

I look forward to being of further assistance.

[TraderLeo]

Hello,

Thank you for the reply.

To further clarify, you said, "I am finding a VERY VERY different behavior of my strategy when back-testing on actual data than on merged data."

Both of these cases would use "actual" data, do you mean that one test using Merged data is different than a test using Non-merged data? If so I would expect this. Could you confirm if this is the case with this question?

I look forward to being of further assistance.

Yes Jesse,
that's the case..
It looks like that the behavior of my strategy on merged data is way different than when I am back-testing on data of the active ES 03-17 (ie. from 12/12/2016 to 2/21/2017) .
Thanks

[NinjaTrader_Jesse]

Hello,

Thank you for confirming.

Yes, in this case, I would expect some differences or even totally different results. The merge policy will directly affect the data you are using. You can see examples of the 3 different modes here: http://ninjatrader.com/support/helpG...ightsub=merged

In the case of using MergeBackAdjusted vs MergeNonBackAdjusted, you would not have the offset applied between contracts which you can see an example of in the help guide. Using DoNotMerge will use data from Only the selected expiry month across the time span of the historical data requested.

Changing the data or period of time in which the strategy runs on can certainly lead to the different result between the two tests.

This would be an item where using Print statements can help see how your logic is executed using the various different merge settings. You can also review what specific data was used by printing each bars data, this could help explain differences you see in the execution of the logic as well.

I look forward to being of further assistance.