I've done some analysis using the most recent polling data on House races from
electoral-vote.com
The outlook is not as rosy as it may seem.
Following the practice of electoral-vote.com, the analysis uses only results from non-partisan polls or the results of the 2004 election if there is no non-partisan poll.
Per electoral-vote.com
Only nonpartisan polls are used; polls taken by pollsters working for one of the parties are intentionally omitted. Unfortunately, this leaves very few polls left, but here is an example of why such a draconian approach is needed. Consider two polls taken in CD-16 in Ohio in which incumbent Bob Ney (R) was being challenged by Zack Space (D).
...
Poll 1: Jun 28-29: Zack Space 46% Bob Ney 35%
Poll 2: Jun 27-29: Zack Space 41% Bob Ney 45%
In poll 1, Space leads by 11%. In poll 2 he trails by 4%. The polls were taken at virtually the same time. The first poll was taken by a firm that works to elect Democrats; the second poll was taken by a firm that works to elect Republicans. We can safely label these, and many others taken by partisan firms as fantasy polls.
There are 209 seats in which Dems are incumbants or lead by more than 8 points (I'm assuming a 4 pt MOE for these polls.)
There are 22 seats in which the polling is within the MOE. In 10 of these the Dems lead, in 8 of them the Repub leads and in 4 (CT-02, CT-04, KY-04, PA-06) there are dueling recent polls, one showing a Dem lead and one showing a Republican lead. In addition, these races generally have a large number of undecideds. In all but 5 of the polls, the # of undecideds is greater than the (assumed) MOE. These races are the ones most vulnerable to any kind of voter fraud or manipulation because they are so up for grabs.
We have to win at least 9 of these vulnerable seats. It is going to be tough.