I hear you and I am familiar with continous contracts. I havent a clue how they would adjust a contango or backwardation on an intraday database that would have relativity for this. Any gaps that would exist between contracts would be errors that could triiger trades and render results useless. Applying the systems to multiple historical individual contract databases would yield correct trades and results and allow correct evaluation. I presented it applied to 8 different contracts above. The only problem is that some of those time periods overlap so even though it is a different contract it isnt really representing totally different market dynamics as contracts from prior years would. I appreciate the comments.
If you study volume a tiny bit, on CL you'll see that the proper time for rollover is at the close, 2 business days before the last trading day. I use the offset between the new contract daily Pivot (H+L+C) and the old contract daily Pivot as my rollover offset. In my experience, it is far better to use merge backadjusted as historical background for the new contract, than that new contract history, because it didn't have enough volume before that rollover to generate "accurate" price-action. If your experience favors using the new contract history, keep doing it ... you still need a couple thousand trades in backtest to start putting any trust in results. Of course, if you feel safe with 49 trades in backtest, start sim-trading that system for a few months (I would use the IB-sim for that, its fills are reasonably realistic).
I would like to run this system on Ninja. I do not have it for systems creation at this time. If someone would like to help me do this, let me know. Thank you