This didn't matter when there weren't enough genuinely competitive states to allow an underdog to "run the table" but now its starting make a difference.

I ran their state by state odds through my model this am while I was running a pot of morning coffee through me. The national variance component has important again, reflecting competitive states that can actually make a difference in the final outcome.

Here's a summary table of 100,000 runs. There is a small amount of noise in the data, this a simulation not a computation. Depending on how word press formats things, this table may come out as gibberish

Nat variance Odds D control

1 .476
.9 .469
.8 .456
.7 .454
.6 .447
.5 .448
.4 .446
.3 .443
.2 .435
.1 .429
0 .402

To get NYT predictions with NYT state odds, the national variance has to be small, in roughly the 0 to .2 range. That implies each state behaves almost independently. If you assume the national variance component is very high, the race much more even. For the record, NYT says "We let the states move together to some extent." That sounds like a small national variance component to me.

It's gonna be a wild ride.. :D

Michale

I tells ya, it's like mid western weather...

I ran the 538 odds table through my own monte carlo simulation (100,000 runs) to examine the effects of local vs national variance on the predicted outcomes. It isn't much, about 2%, with a larger local variance component favoring the Dems. With 100% local variance my outputs shows Dems retaining control of the senate 46.3% of the time. Nate's call was 46.2%, suggesting he weights the national variance towards 100% of the total. As of today, NYT 43%, WAPO 49.9% and HUFFPO 48% chance of Dems controlling the senate.

My quick and dirty rank ordered model assuming 100% national variance is matching up very well with Nate's odds tables lately. Most recently, his 46.2% to my 47%.

With the latest 538 numbers, Kansas seems the tipping point for Dems, with Iowa and the-hard-to poll-Alaska not far behind. If any one breaks strongly towards the Dems, Blue starts to look like a 60:40 favorite.

There is still a lot of volatility in the polls, but any way you slice and dice it, this cycle looks to be shaping up as a nail biter.

Michale

Typical politician...

"I am going to be with the Party that's the winner"

That attitude is likely going to turn off Kansas voters... Who wants to vote for a representative who won't state upfront his views??

Piddle or get off the pot...

Michale