While he's smuggling in a lot of conservative talking points I disagree with into his political analysis, I do agree with Sean Trende's overall contention that an 80 or 90 seat GOP pickup this November is well within the realm of possibility. (Trende also concedes that an economic recovery and a bump for Obama could result in a mere 20-25 seat loss.)
Trende also makes a point that's gotten too little attention: the apportionment of the House gives Republicans an inherent structural advantage:
The President's weakness in these states reveals another problem for his party. Since he is weak in Republican areas and swing areas, and yet doesn't have horrible approval ratings overall, he must be very, very popular among his party's base. Some polls have his approval ratings among African Americans at 95%. Even in Massachusetts, Martha Coakley managed to win the First, Seventh and Eighth Districts, which are home to the state's liberals and minorities.
The problem for the Democrats is that these voters are packed into a relatively few states and Congressional districts nationwide, diluting their vote share. This is why the median Congressional district is an R+2 district. Thus, the President could have a relatively healthy overall approval rating, but still be fairly unpopular in swing states and districts. The increased enthusiasm that Obama generated among minorities, the young and the liberal is useful, but only if it is realized in conjunction with Democratic approval in a few other categories.
The House is less tilted toward Republicans than the Senate, but it is tilted. Democrats "waste" a lot more votes and voters in non-competitive urban districts. This is one reason why Republicans have in the past been able to control the House despite a majority of polls showing them losing the generic ballot. If the generic ballot is tied among likely voters, Republicans will probably gain a House majority.