I'm now thinking about how you would run an election-based RTD game.
Let's suppose a given American presidential election from real life, say, Nixon vs. Kennedy. Players are randomly assigned states, but these are kept secret except to the GM. Players can "vote" for a particular candidate, and a six-sided die is rolled, with the usual epic fail-fail-partial success etc. metric to determine how much the margin of vote in that state deviates from real history.
For example, let's say we were RTDing the 1960 election, and you cast a vote for Nixon. A random number generator gives assigns you Texas, where the vote margin was 2% in Kennedy's favor. I roll a 5 (Epic Success), which switches 5% of the vote (115,000 voters) to Nixon- who now carries the state by a 3% margin.
Assuming I somehow rolled a 2 on everyone else's states and nothing changed from real life, Kennedy still wins the Electoral college, but loses the popular vote.
I'm not sure how you would do primaries, and once you get into the butterfly effect, you no longer have real-life matches to use as baselines. How, for example, would you determine a baseline vote for a 1964 Kennedy vs. Goldwater match? It can't be the same as the actual 1964 election, because part of Johnson's success was sympathy voting from Kennedy's assassination. Kennedy would probably win, but how much by?
It's an intriguing idea, though, I'll admit...
[EDIT: You might be able to use the Economist's "Misery Index", which is the sum of the inflation and unemployment rates, to simulate the effect of the economy on elections. We could say that a "neutral" economy that neither helps nor hurts an incumbent president has, say, 3% inflation and 6% unemployment, giving 9 "misery points", and every two misery points is a point knocked off the incumbent's margin.
Which works well if you're simulating normal incumbents. But this also predicts that Roosevelt should have been up against a 9-point margin against him in 1936, which is obviously not the case...maybe if you looked at the change in misery rate between the start and end of a term?
(In that case we clearly need to rate unemployment as more dangerous than inflation to an incumbent's record, because there was a 10% deflation rate in 1932, but a 1.5% inflation rate in 1936- that means their misery rates balance out to about the same.)]