The dude seems like hes way overcomplicating it.
At some point, the numbers have to be assigned to a given precinct.
They picked red counties, and just nudged the votes enough to take a few points off Trumps lead. Or picked safe blue counties (with friendly election boards) and nudged it even more.
The goal was to get within striking range, but not trigger statistical alarm bells.
Then they could inject whatever else they needed into the big cities with corrupt boards.
I think the ratios are just artifacts of trying to goal seek a certain percentage that is within 3% of historical averages.