The last time Americans went to the polls in substantial numbers, two clear winners emerged. One was the Democrats, and in particular Barack Obama, who won his second term and secured an increased majority for his party in the Senate on his coattails.

The other winner was election forecasting — and in particular its most prominent exponent, Nate Silver of FiveThirtyEight, who left his many critics in the conservative media looking exceedingly foolish when his model proved accurate to the last state.

Two years on, each faces a considerably more difficult environment as Americans — or at least the 40% or so who seem likely to make the effort — prepare to vote in Tuesday’s midterm elections. Along with sundry governorships, state legislatures and other offices, voters will be choosing a new House of Representatives to serve for the last two years of Obama’s presidency, along with 36 of the seats in the 100-member Senate.

Much as byelections do in Australia, midterm elections grant disgruntled voters an opportunity to take a swing at the government without having to consider too carefully the quality of the alternative. But whereas byelections are usually purely symbolic in effect, the midterms very often leave a lame-duck president pitted against a powerful and hostile Congress, and are a major contributor to the gridlock that blights the American system of government.

After gaining control of the House of Representatives in 2010, the Republicans were able to retain it in 2012 despite the defeat of Mitt Romney’s presidential bid on the same day. For this they can partly thank gerrymandering, although the Democrats are also handicapped by the inefficient distribution of their voter base, too much of which is locked up in low-income urban districts. With the swing sure to be in their favour this time, there is no prospect of the Republicans’ grip on the House being loosened tomorrow.

The real game is the Senate, in which the Republicans need a net gain of six for an absolute majority of 51 seats, as they would require to break a casting vote held by the Democrat Vice-President Joe Biden (and bearing in mind that much of the work of obstruction can be accomplished with the 40 votes needed to sustain a filibuster).

Going back to the end of World War II, the average midterm election has cost the party of the White House incumbent four seats in the Senate, becoming more like five or six when the president has been in his second term. The only second-term president to hold his ground through this period was Bill Clinton in 1998, a particularly interesting result given he had spent the year embroiled in the Monica Lewinsky scandal.

That means the Republicans’ performance will need to be at the high end of average to get them over the line. They would have good cause to feel optimistic about such a result, given the extent to which the Democratic triumph of 2012 was built on Obama’s capacity to draw traditional non-voters to the polls.

In attempting to ascertain the likelihood of a Republican breakthrough, we could of course do a lot worse than refer to the forecasters who hit the nail on the head in 2012. However, what they’re telling us is that we shouldn’t expect as much from them this time, because the midterms produce a relatively small volume of lower-quality polling.

Even so, the clear and growing consensus is that a Republican triumph is considerably more likely than not. It is agreed that seats in Montana, South Dakota and West Virginia are as good as won, each being a Red state in which a Democratic incumbent is retiring. The Republicans’ hopes for at least three more rest upon Alaska, Arkansas, Colorado, Iowa, Louisiana, New Hampshire and North Carolina.

Conversely, the Democrats hope to gain Georgia from a retiring Republican, and a Republican incumbent in Kansas faces a strong challenge from an independent who might decide to caucus with the Democrats under the right circumstances.

The six major forecast models currently put the Republicans’ chances at between 60% and 94% respectively in the case of the Princeton Election Consortium and The Washington Post’s Election Lab.

While each has the Republicans rated considerably better than even, most allow for a non-trivial chance that the Democrats might just pull off a surprise. However, this uncertainty cuts both ways, with an equal probability that the Democratic defeat will turn into a rout.

In this respect, the most recent portents for the Democrats have not been good. Of the 12 polls published from battleground states over the weekend, only one from North Carolina and another from Iowa had the Democrats in front, and then only by the barest of margins. And just as they did in 2012, the modellers’ forecast probabilities are being revised upwards in favour of the front-runner with every day the election draws nearer.