With all due respect and seriousness.
How about testing and carrying over that same test into actual play?
What I mean is, doing the theory test with XYZ number of shoes/hands, 10,000 or 100,000 or 1,000,000, whatever the appropriate amount of total numbers are.
Then carrying over that same number, divided by say the, 1 or 3 or 5 shoes that someone will actually play in a casino with the number of hands the player would actually wager on.
I say that because to me, results from theory of testing are correct. But, a player has to be able to replicate that when he sits down to play. And the player is not going to play 60,000 or 200,000 hands.
That should give you the actually wins or losses of what the person would have realized, no?