Modelling Quality in back-testing

 
Hello guys,

I am confused so asking here, 

which modelling in back testing method is more accurate?

A) Every Tick or
B) Every tick based on real ticks?

I tested with same input parameter, however, getting absolutely different results in both?

Please suggest me.
 
'Every tick based on real ticks' modelling is the best, because your MT5 strategy tester download real historical ticks from the broker's server.