Much has been made of the lack of a Democrat bounce after last week's convention. I actually discount all the polls I hear about, though I admit I've got to do that as an act of the will - it's easy to be suckered into believing there's actual substance to them.
Now, investing - or what some would call gambling, as there are no actual earnings involved here - is another matter. This represents "polling" done by people who are willing to back up their opinion with their cash.
Here are graphs of several races as tracked by the Iowa Electronic Markets:
2004 Presidential Election, based on who wins the popular, not electoral, vote.
2004 US House Control, where RH_Gain represents Republicans gaining seats, RH_Hold represents Republicans retaining majority without gaining seats, and RH_Lose represents Republicans losing majority status altogether.
And my favorite graph of all, except for the unfortunate fact that neither of New Jersey's Senators are not up for reelection this year:
2004 US Senate Control, with similar definitions to the above, except that RS_lose also includes the case of a 50-50 split between Republicans and non-Republicans.
Near the end of the Democrat's convention, Larry Kudlow, at National Review Online cited these data, noting at the time (using earlier data, about July 25) that the Senate data were in favor of RS_Lose. Seems that any concern about the Democrats winning the Senate started evaporating on July 28, the date of John Edward's speech.