At this point market and polling data suggest that a change in House control is significantly greater than continued Democrat control.
While it is always dangerous to speculate, traders are likely to upgrade House Republicans chances over the next week should current generic ballot trends continue. Democrats are quickly approaching the equivalent of needing to draw an inside straight in order to hold onto the House.
The Senate still poses significant challenges for Republicans (from a control perspective) but as I mentioned this morning the path to control is becoming more defined. Senate control could very well come down to whether or not Democrats can hold one of three seats (Illinois, Nevada, and Washington). It's still too early to declare this as the Democrats "three seat firewall" but if the situation remains stable over the next two weeks, that is where we are likely to be a week out from the election.
The Iowa Electronic Markets-- price the probability of a House Republican takeover at a slightly higher level of 78%. Those markets also price a Republican Senate takeover slightly higher (22%). There is no clear cut reason that easily explains the the House discrepancy, but the Senate difference likely results from the fact that IEM counts independents caucusing with Republicans or Democrats toward determining control. InTrade does not. In the case of the Senate the best example of this difference is how a Lisa Murkowski victory would be treated. Under IEM she could be the 51st member counted toward control. That is not the case under the InTrade market contracts. Lieberman and Sanders count as Democrats for control purposes.
A word on poll averages ... while this is an extreme case when looking at poll averages observers should put a plus or minus 1% "uncertainty margin" on the reported spread. Why? Because polls are generally not reported with raw numbers, but with percentages rounded to the full percent. The effect is that a margin of 45 - 44 could range from 44.5 - 44.4 (0.1 spread) to 45.4 - 43.5 (1.9 %) spread. While the averages probably tend to cancel each other out, it is worth remembering when looking at narrow spreads in poll averages.
