currently i have EOD historical data, so i don't have 15 minutes to check backtests so i will do everything using 1 DTE without assignments although i could be assigned before the 1 DTE, but i am hedged (spread) so it will probably be ok, right ?) thanks.
IMO: When I create some back-test, I try to insure the back-test will closely approximate a real trade. So, for the case you mention, for taking the trade to expiration, the "real trade" would be to close in the last 15 minutes before expiration, which can approximate a "cash settlement", and should be very close in pricing as the simpler back-test algorithm, which would actually take it to expiration. This solves my desire to have the back-test fairly accurately represent a real trade. For your model for the closing price, you use the closing price of the Underlying, which you compare to the Strike price of your options (the EOD value for the option may have little value -- You would use the EOD value for the underlying.). For any expiring option that is 1 penny or more ITM, assume assignment with the cost set to the Strike price. So, you can back-test either scenario (1 DTE to close, close at Expiration, or allow expiration and close any assignments later), your choice, but keep in mind if a "real trade" inferred by your back-test results, is possibly something you may never consider doing (inferring that some of your effort may not benefit you) Regarding "Early Assignment", YES this is possible, and yes, you have the other leg of the spread as protection.
99% of option traders have no intention of trading the underlying. The idea behind an IC is to let the legs expire worthless. Assignment of a ITM option is NOT AN OPTION.