Backtesting Results questionable

[kenb2004]

I've been running backtest results on CL and I noticed when backtesting the same strategy over a 2 week period that results from CL-04-11 were indentical in every way to the results of CL-05-11 through CL-11-11. How can this possibly be accurate?? Data series is 1 second with Default 24/7 setting.
thanks

[NinjaTrader_Matthew]

Hello kenb2004,

Thank you for your post.

This is likely due to the merge policy settings you are using on the CL instrument when backtesting.

You can avoid this issue by going to Tools--> Instrument Manager--> Locate the CL instrument and select "Edit".

From the Instrument Editor, locate the Merge Policy settings and set this to "DoNotMerge" or "MergeNonBackAdjusted".

You should then see the CL 05-11 contract with the correct last price, rather than backadjusted to match the current prices of the April Contract.

You can read more about Understanding Merge Policy settings from the following link:

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

Please let me know if there is anything else I can do for you.

[kenb2004]

Thanks, but what is the difference between "DoNotMerge" or "MergeNonBackAdjusted"? Which is most appropriate for this case?

[NinjaTrader_Matthew]

kenb2004,

DoNotMerge is only going to show the selected contract without merging into the previous contract. This means if you were testing 2 weeks of data, you might not get your full historical data request for the 05-11 contract.

MergeNonBackAdjusted will merge the historical data, however will not use an offset as the default MergeBackAdjusted does where you were seeing the same data on two contracts.

If you would like backtest each individual contract, I would suggest DoNotMerge. However, if you need 2 weeks of data on the May contract, you will need to use MergeNonBackAdjusted to receive this data.

[kenb2004]

Thanks for your info

[kenb2004]

I made changes to "MergeNonBackAdjusted" in both Instrument>CL>edit and Options>Data>merge policy and backtested with the same identical results. What else do I need to do?

[NinjaTrader_Matthew]

kenb2004,

Thanks for the update.

What happens if you use DoNotMerge?

[kenb2004]

When I use DoNotMerge they are unique.

[NinjaTrader_Matthew]

kenb2004,

This is due to the 04-11 contract still trading and it is merging the data into the 05-11 contract, which yields very similar results.

If you need to test for unique results per contract, please use the DoNotMerge function until the 04-11 contract settles.

[kenb2004]

So why did it continue to merge everything when I told it MergeNonBackAdjusted?

[NinjaTrader_Matthew]

kenb2004,

This option will still merge everything, but it will not try to 'back adjust' the the data to an offset value from the previous contracts.

By default, the MergeBackAdjusted option will try to combine the two contracts and adjust the values of the previous contract to match with the current.

MergeNonBackAdjusted will still merge the historical data, but it will not adjust to match. As a result, you may see gaps in the data. However, since you are seeing the same results with both of these Merge settings, your strategy must not be sensitive to this offset value and will yield similar results.

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

Please let me know if you need further clarification.