I don't assume Biden over-performed at all. In fact, he didn't, which you can see in the post I did with real data here.
But you called into question the validity of that data, which I agree with. Which is why I turned to simulations. Even in the case where Trump over-performs, you still get a "regular line", as Shiva says. Remember he claims that any deviation from flatness is a sign of fraud--which is patently false, since a line of some sort is the likeliest result when plotting the quantities he's decided to use.
I don't think I'm being misleading since I say as much several times. If the "true slope" of the correlation between split ticket Trump voters and straight ticket Repubs is greater than 1, you'll get a line going up and to the right. It's still a line, which Ayyadurai claims is anathema, in his video. But it's not. It's expected.