This would be an interesting question even if it lacked political context, but the electoral context adds even more:
Generically, the Democrats have a numerical advantage over the Republicans in U.S: There are more potential Democrats, but their turnout is more variable. The only paths that the Republicans have for winning the race is to gamble, and there are two ways to gamble.
First, the Republicans might gamble on the Democrats getting unlucky: they might hope that their voters turn out with the usual reliability, but somehow, Democrats get unlucky and their voters don’t turn out. This was basically Mitt Romney’s approach in 2012, and this was not as unjustified in retrospect as might have seemed then: the midterm elections of 2010 and 2014, if anything, underscore how unreliable the Democratic electorate is. But, for 2012, with Barack Obama at the top of the ticket generating more reliable turnout among the less reliable segments of the Democratic voters, they were probably going to be as reliable as possible, and they were exactly that in practice.
Second, the Republicans might gamble on their own “lucky” shot. Instead of getting fewer votes with certainty, they might change their approach and gamble on a chance of getting more votes, with the downside of falling apart completely. This is what Donald Trump’s campaign is all about: he runs a respectable chance of outperforming Romney among the working class whites and is risking a big fat chance of losing a lot of the regular, reliable Republican votes–and he’s gambling that he won’t. So he will probably do much worse than Romney (in terms of the mean) but has a decent chance that he would do better. But, again, the Republicans are facing the wrong candidate. Unlike Barack Obama, Hillary Clinton is about as most unreliable vote getter among possible Democratic candidates as possible: she will get most of the establishment Democrats, but enthusiasm for her among the liberals, the youth, ethnic minorities, etc. is low. They will not vote Republican…but the real risk is that they will vote third party if they vote at all–posing a significant risk for the Democrats in lost votes. This is exactly the scenario where the reliable but few regular Republican electorate probably stood an excellent chance. Instead, they are gambling with a Trump.
As a generic game, the game of war (the card game) with uncertainty, featuring trade off between mean and the variance, does not really have a clear solution: it is really just a variation on the matching pennies game. Of course, the parties did not necessarily choose their candidates “strategically.” Still, it is fascinating how mismatched the Republican and Democratic candidates were and still are, in 2012 and again in 2016.