You are somewhat unfair to NT backtesting abilities ... it is always possible to submit your orders to a 1-tick or 1-sec timeframe (I prefer the later, less memory requirements and actually better simulation of transmission delay to the exchange), which makes your backtests results as accurate as it gets - provided you are using CalculateOnBarClose=true (which should always be the case, in my professional programming view).
If you need intrabar resolution, without falling into the evil CalculateOnBarClose=false, you can either use a lower timeframe (say, using 1-sec TF instead of 1-min TF), or use a 1-tick timeframe and reconstruct your main timeframe bars from it.
In my experience, a 1-sec TF supplementing your main TF is enough for most intrabar trading decisions.