I believe as much as anyone that this election is highly suspect, but there is no “smoking gun” here, just atrociously bad analysis. Essentially, he sorted the data in a way that’s basically guaranteed to give him the result he is then “surprised” to find.

First, let’s consider how he prepares the data for analysis (see video starting from 1:59): He used data only from Philadelphia, which is a heavily-Democrat city. He then sorts this data by “Trump Votes / Total Votes” (which he calls the “major ratio”) as the primary sort key and then by timestamp as the secondary sort key.

Then (at 3:44), he seems surprised that you have to go pretty far down in this data to get ....on and on.

Can you just wrap it up into a simple form for a simpleton?