yes, I'm too find that this calculation is too much rounded and could be much precise.
Authors say that the calculations based on following study http://www.hiddenloft.com/notes/SubExposure.pdf
Ideal Exposure Time = (9.76 x ReadNoise^2) / SkyFlux
SkyFlux = (BackgroundADU x Gain) / TestExposure(Min)
We can see now that SGPro takes all of the parameters from what we have provided him and from test exposure.
The problem is I believe, that the BackgroundADU should be taken from nebula/star free region in the test frame and more important dark-substracted, which doesn't happen inside SGPro. Seems like it just takes Median value of the test frame and applies calculations.
I can only suggest to implement an option to enter our Bias frame median value, (profile specific) as we do with gain and readnoise and use this formula:
Ideal Exposure ADU = 9.76 x (r^2 / Gain) + Bias Frame Median ADU
With this you don't even have to take test exposure frame, after inserting all the details, SGPro can show you your desired median ADU value that you need to reach, so you'll adjust time accordingly by your self.
Giving it a second thought, it can also be used with taking test frame.
Took 3 minutes frame, noted median ADU and from that suggested corrected time needed to reach desired median ADU.