Political scientist Robert Fair of Yale University has an uncannily accurate track record in predicting the outcome of Presidential and Congressional elections, based solely on various economic factors. In 2008, for example, he predicted the Democrats would receive 55.8% of the two-party vote for the House, and on election day the Democratic share was 54.8.
This weekend, Fair made his final projection for 2010:
The current forecast (dated October 29, 2010) of the Democratic share of the two-party House vote is 49.22 percent. This has changed only slightly since last October. The equation has always predicted a close election. The current forecast is based on actual (preliminary) economic values, values that were released by the BEA on October 29, 2010. INFLATIONCC is 0.97 percent, and GOODNEWSCC is 1. The low inflation is good for the Democratic party, but the fact that there is only one good news quarter is not. On net, the Democrats are predicted to get slightly less than half of the two-party vote, 49.22 percent. This is down from 53.4 percent in 2008 for the on-term House vote.
I do not have an equation that translates the vote share into House seats, and so I have no prediction of House seats
By my calculations, this means that the Democratic share of the vote for the House is 89.8% of what it was in 2008. In 2008, 54.8% of the two party vote was good enough for 257 seats. If the Democrats win 89.8% as many votes in 2010 as they did in 2008, I have no idea how that translates to losses. Just as a straight calculation, 435 x .4922 = 214, which would be a loss of 41 from the current 255. But your guess is as good as mine. If Prof. Fair has overestimated the Democratic vote by 1 percentage point, than 435 x .4822 =210, a loss of 45. In the current atmosphere, I would almost find a loss of "only" 45 seats a good outcome.
You can find the link to Fair's data here.