I’ve been looking at which party is going to control the Senate after the election, and I’m a bit surprised at some of the juxtapositions.
The Friday prices on the Iowa Electronic Markets show a 83.5% chance of Democrats gaining control of the House and a 33% chance of gaining control of the Senate (i.e., the Republicans having fewer than 50 Senate seats).
The Saturday evening TradeSports line is similar: the Democrats have an 87-88% chance of taking the House and a 31.6-31.9% chance of taking the Senate.
The House races are too numerous for me to make much sense of, but as I read RealClearPolitics, the Democrats need to pick up 6 seats to control the Senate, assuming that the probable 2 independents (Lieberman and Jeffords[‘s independent replacement Bernie Sanders]) caucus with them.
Likely pickups are PA (Casey +11.2%) and OH (Brown +11.2%).
Leaning Democratic is RI (Whitehouse +9.2%).
The other three possible pickups are listed as tossups (MO, MT, and VA), but the Democrat is ahead in the RCP poll averages in each race by 0.6% to 1.7%.
The Saturday night TradeSports bid/ask spreads also support Democratic victories in each individual race:
MO: DEM: 50.3% – 55%; REPUB: 46.1-48%.
MT: DEM: 65% – 68%; REPUB: 35-39.8%.
VA: DEM: 55.2% – 58%; REPUB: 43.3-48.2%.
So if each individual race breaks as currently polled, the Democrats win the Senate.
There is an outside chance that the Democrats could pick up one of the other two Republican seats that RCP lists as leaning Republican (AZ, 8%, and TN, 6.5%), even if it lost one of the seats listed above.
If each race individually goes according to the current polls and current betting line, the Democrats win both houses of Congress. But since so many races have to go toward the Democrats for them to win control of the Senate, both TradeSports and the Iowa Electronic Markets reflect about a 2 to 1 odds of the Republicans keeping the Senate.
I find this an interesting example illustrating joint probabilities: combining even highly correlated multiple probabilities more than 50% yields a joint probability significantly less than 50%.