I think there have been presidential elections where the winning party was going to win, baring some outside event, that neither side had control over, breaking against them.
There are other elections where the winning party would win unless they screwed it up really badly. And that is without getting into the fixed elections.
I'm not sure how a historical election simulation game would handle the issue of fixed elections. I think at a minimum it has to be dealt with in a 1876 scenario, since the only reason the election is known at all is because it was fixed. Also 1864, since regardless of what Lincoln may have put in his letters, the GOP was simply not going to let Lincoln lose that one, though they might have allowed a McClellan win without the Copperhead plank and a different running mate.
In the specific case of 1984, how to make it competitive has been debated on this forum. Assuming the first Reagan term goes IOTL, and the game scenario tends to assume these things, Reagan as running as a Republican incumbent president during the Cold War, when incumbent Presidents who were not thought "weak" (Carter) would win, and often win big, and Republicans tended to win as well. And the economy was as good as it could have gotten, though people thought the Democrats had a chance earlier in the year, when the economy was in a worse condition. So with these parameters, Reagan is going to win and the issue with playing Mondale is if he can do anything to make it closer.
So for a situation like this, you need a random event where Reagan's dementia is both worse and more obvious than IOTL, which with a POD after March 1984 was the only thing that was going to stop him. This could happen in maybe a fifth of the games. If this happens when playing as Reagan, the player would have the option of replacing Reagan with another presidential candidate (the Vice Presidential candidate doesn't automatically replace a Presidential nominee who withdrawals, though in this case the Republican National Convention would almost certainly have reconvened and nominated Bush and a new vice presidential candidate), or toughing it out. When playing Mondale, when this event occurs, sometimes there is a new Republican candidate and sometimes there isn't. Otherwise you are trying to do better than Mondale's historical result though there is another issue that the only real break that campaign got historically was the first debate, everything else broke against it.
And since some player made scenarios explore different nominees, there could always be a scenario where the race is between Bush and Mondale, where the Republicans would have the advantage, but Reagan's charisma is removed. Reagan opting for a single term due to worse health or his wife's wishes is a plausible POD.
The 1956 election is similar, there is one wildcard, which is Eisenhower's health, and that is it. You have to have Eisenhower withdraw to make it competitive. I don't think you can even do this for 1952, which was actually not a blowout in the final results, but its hard to see a path for victory for Stevenson. With 1972 you have to have Watergate and other scandals blow open in a way that the Nixon player can't get control of. With 1964 there is nothing you can do to give Goldwater a chance except to have LBJ plan the JFK assassination and get caught, which other than the CT is still ruled out as being an event that occurs before the scenario.
To look at other postwar elections, the incumbent is probably going to win in 1996 and 2012, and lose in 1980, but in these cases if the player playing Dole/ Romney/ Carter makes a lot of good decisions, the random of the algorithm might push him over the top. I don't know how the algorithm works.
Before World War 2, there wasn't the Cold War incumbent advantage, but there were cycles of dominant and out parties and periods where things were pretty stacked in favor of the dominant and against the out party.