I'd say it's only stupid for the people who bet on Grimes, because if they want to bet then they should be allowed to take the 4:1 or even 8:1 favorable odds based on disengaged/clueless people contending that McConnell is upwards of 80%-90% to be reelected and then cash in huge in November because Grimes has an incredibly strong chance at winning based on current fundamentals. If people want to place their faith in a projection so obviously at odds with the data then the Grimes' betters should insist on putting up only 25fg for every 100fg or 200fg that the McConnell better puts up, and then they should smartly bet as high as the McConnell better is willing to go.

Some serious shit has already gone down, that's why the race has been locked in a statistical tie for a year despite the numerous advantages that McConnell brought into the race that are yet to really work to his favor. McConnell has to bank that some huge consequential event is going to change the dynamics of this race because we're on a collision course to a 49%-46%-5% win for somebody and it's just as likely to be Grimes as it is McConnell. Right now Senate control is also 50%-50% (largely because this race is 50%-50%) but we are right on the knife's edge though, meaning the slightest move in either direction could push the takeover likelihood up to 80% or as low as 20%.

I think I'd take Grimes in 8:1 odds. 4:1, not so sure.

You should absolutely take Grimes at 4:1 odds.

One of the biggest factors in this race that is yet to manifest itself because it's still so early is that McConnell doesn't overperform his polls. His performance in that regard relative to other incumbents is very, very bad. It's one side-effect of his low favorability and job approval. He is very lucky to outperform his polls by 1-3 points and that's only in a 2-person election which this will not be because there will be a Libertarian on the ballot.

It will no doubt have an impact on this race, how much will be determined by the quality of polling after Labor Day. If we could run a controlled simulation that held that no large external events impacted the race in either direction between now and Election Day, the polling would move in the direction of the Democrats almost automatically because more nonpartisan and media polls will be introduced into the field and they will drastically lower the share of polling that is currently being eaten up by polls that are either questionable or outright bad. That is going to impact the race (if the polling is even semi-accurate) because it will either lower McConnell's share or it will raise Grimes' share. It might even do both.

Long story short: if we're in mid-October and McConnell is polling at 45% or 46% then he is very likely done. If he's polling at 47% or 48% then the race is a tossup. McConnell truly needs to poll at 50% in an aggregate of accurate polls in order to feel like a favorite.

It's difficult for me to take those odds because of how Kentucky votes for Senators and Presidents. When is the last time Kentucky went to the Democrats? Sure, you have Southern Democrats in Congress and state offices, but Kentucky seems very....red when it comes to the state as a whole.

The Democrats last carried the state in a presidential election in 1996 and in a Senate election in 1992, but that doesn't speak to the competitiveness of the races especially in the Senate. Senate candidates have badly Republican presidential candidates even when they themselves are incumbents. McCain won Kentucky by 16 while McConnell won by less than 6, and in 2004 Bush won by 20 points while Jim Bunning won by only 1.5. The record during midterm elections isn't better or worse: in 1998 Bunning won by 0.6 but in 2010 Rand Paul won by 11. The sample size is incredibly small and obviously tainted with respect to that 2010 figure.

This is caused primarily by a large contingent of voters who still back Democratic candidates on occasions and who refuse to vote for McConnell or Bunning, as evidenced by their several close-calls. The PVI of a state is only right until it isn't any more. For example in North Dakota in 2012, which is about as close of a comparison as you can find to this race, Heidi Heitkamp won in an R+10 state more inhospitable to Democrats running for federal office even as Romney won the state by 20 points. Ultimately these races come down to the candidates, their strength, their resources, and their strategy. McConnell rarely outperforms his polls and when he does it's by fractions, and in 2008 he never polled under 47%-48% outside of one obvious outlier. Contrast that to this year where McConnell is not even regularly hitting 47%, and can only do so in stray surveys from Republican internals and the worst-performing known firm in the industry (Rasmussen). There is a big, big difference between this race and past Senate elections in Kentucy.

New Ranges, shift, and cause for movement.

House: D+3 - R+10 - [Shift: <-] (Results of GOP primary election in WI-06)
Senate: R+1 - R+7 - [Shift: No Change]
Governors: D+6 - R+1 - [Shift: No Change]

The outcomes of tonight's primary elections were pretty predictable but particulars of those elections bring some good news for Democrats. The AP has called the primary election in Wisconsin's 6th district for State Senator Glenn Grothman. This R+5 district went 53% for Romney and was generally thought to be Safely Republican unless Grothman pulled out the victory tonight. He has a long list of very questionable statements that both Democrats and Republicans have attacked and him winning the GOP's nomination puts the seat in play for the Democrats.

While there's no change in the Governor's ranges, national Republicans are probably disappointed that winner Tom Foley currently holds an underwhelming 56%-44% lead despite having been the party's de-facto nominee since his narrow loss four years ago. This is actually one of the only two or three real pickup opportunities for Republicans in November so if this showing is a reflection of how competitive Foley will be then the net gain in November will be much closer to the Democratic edge of the range, D+6, than the Republican edge of the range, R+1.

The fundamentals of the race favor him, despite his personal weaknesses. I don't see a lot of Republicans or Republican Independents peeling off when the Senate majority is within their grasp, and this has been a very good year so far for Republican Senatorial incumbents (re: their teaparty challengers).

One quick point up front: this has actually been a terrible year for Republican Senate incumbents. None have lost but John Cornyn, Lindsey Graham, Thad Cochran, Pat Roberts, Lamar Alexander, and yes Mitch McConnell all set records in their states for the lowest primary vote share ever recorded by a sitting Republican Senator. If people are counting on reflexively-Republican voters to show up and bail out McConnell or to save Georgia or to get them over the finish line in Louisiana or Arkansas or Alaska then they're screwed. Straight up.

The fundamentals of the race don't favor McConnell, they don't favor anybody. McConnell has an advantage in some but a deficit in others, and some of his deficits are large if not outright disqualifying.The voters you described don't behave that way ("true Independents"/Rank 4 partisans anyway) and they have no expectation that their vote will impact the balance of power either way. The last time he was on the ballot those voters were pitched the "60th vote" scenario and they didn't care -- they didn't step up to save McConnell, he barely won, and he barely outperformed his polls. Even if there were enough of those voters persuadable to the argument, there's nothing that suggests they'd side with McConnell the way the race has progressed this far, the way they feel about him historically, and the way they feel about the GOP.

If this election were in Georgia or Louisiana or even Alaska then some voters could be persuaded by the type of argument you outlined, because they would be making the decision on their own. Should either Georgia or Louisiana go to a runoff (or both) with the Democrats already winning 49 seats the race(s) will be framed in that exact fashion: vote for Mary Landrieu to keep her Chair of the Energy committee or vote for Bill Cassidy to give Republicans a majority in the Senate! This race isn't that, though.

This is what pundits are counting on. The polls show a tight race, some show McConnell slightly ahead. The general view, whether right or wrong, is that he will scrape out a win come November.

Not a chance. The more notable pundits haven't even begun to discuss that important aspect of the race (I doubt most of them are even aware of it given its absence) and if they were it would point them to the opposite conclusion: that McConnell narrowly loses in November. If McConnell barely outperforms his current standing in a robust aggregate then he loses in November no questions asked. He would either need to: 1) break the race wide open, 2) raise his standing in the polling and then match it as he's done historically, or 3) greatly overperform his standing in polling having never done that before in his history of a candidate, even when he was more popular and facing weaker opponents.

The pundits in the aggregate are counting on "midterm election in Kentucky during 6th year of Democratic presidency" and nothing else, even when the available evidence is telling them that McConnell is in deep trouble and even needs some assistance to win. There isn't a general view of the race either: 538's preliminary model has McConnell at an 80% chance to win, the Upshot has McConnell at a 77% chance to win, the Monkey Cage has McConnell at a hilarious 98% chance to win, and Cook/Rothenberg/Sabato rate the race Tossup/Lean R/Likely R respectively. Taken together the pundits tell us the race is either a tie, slightly favors McConnell, or McConnell is a lock to win. Exemplary analysis from people paid to evaluate elections but who appear to be paying very little, if any, attention to elections.

There is positively no reason to believe that McConnell scraping out a win is any more likely than any other outcome. At best people are adhering to outdated conventional wisdom rather than to the data, and at worst people are outright ignoring the data because it doesn't comport to what they want to believe about the race.