just a random idea i had, what if Presidential elections were held in the environment of the preceding midterm. Basically what I did was just take the Presidential approval rating on election day irl and compare that to their vote percentage, then apply that difference to their approval rating in the preceding midterm and swing the results by the difference between that and their actual vote % nationally. some interesting stuff.
1984 held in 1982:
Reagan still wins, but it's much closer.
1992 in 1990:
I assumed that Perot took equally from both Bush and Clinton for this one.
1996 in 1994:
Bob Dole actually pulls it off.
2004 in 2002:
Bush landslide
2012 in 2010:
Romney wins, more narrowly than I was expecting though.
2020 in 2018:
Trump additionally loses Florida and North Carolina. Surprised actually, I feel like a Trump victory would have been more likely in 2018, not less.