One Thing It's Hard for Election Models to Account For ...

... is the conflicting strategies of the Democrats and Republicans.

The Democratic strategy is to GET OUT THE VOTE -- even if it that means getting it out of graveyards, etc. In a close race, if the Democrat is behind, count on more ballots to magically appear.

The Republican strategy is to make it as difficult as possible for poor people, black people, etc., to vote. Then, in a close race, if the Republican is behind, sue left and right to stop votes from being counted.

Here's how my model accounts for that:

Nationally and in competitive areas, I consider it a wash. The Republicans are probably going to subtract as many votes as the Democrats add.

Locally, i.e. in state by state prediction, I look at which party is in power, where the political machines are and how powerful they are, etc. close the state looks, and assume  a 1% edge in addition to what polling might indicate on the part of the party with better fuckery machinery (unless one of their schemes is visibly kiboshed, in which case I deduct that advantage).