(CNN) First things first: The theme song of the week is "Nothing's Gonna Stop Me Now" by Jesse Frederick and Bennett Salvay from the television show "Perfect Strangers."
Poll of the week: A new Marist College poll finds that Democratic Sen. Bill Nelson holds a 48% to 45% advantage over Republican Rick Scott.
What's the point: For much of the summer, Scott has been tied or ahead of Nelson. Some of that polling was from lower quality pollsters (who don't call cell phones or don't use live interviews), but even among higher quality pollsters it seemed Scott was more than holding his own.
A Scott win would be against what we'd normally expect to happen in a midterm election in which there is a Republican president. In midterm elections since 1982, the senator of the opposition party has won about 96% of the time. The only cases where that didn't happen were when the president was extremely popular (President Donald Trump has an approval of around 40%) or when the incumbent was suffering from a scandal (Nelson isn't).
Indeed, I'm not aware of a "fundamentals" model in which Scott would be favored. My own model that takes into account money raised from individuals, candidate quality (i.e., current or previous office held), the partisan lean of the state and the national political environment has Nelson favored by a little less than 10 percentage points.
Put another way, Scott has been doing better than the fundamentals has suggested he would do. The average poll still finds this but to a lesser extent. That is, the polling has moved towards the fundamentals.
That the polling seems to be moving towards the fundamentals is good news for Democrats not just in Florida but nationally as well.
The reason is that the fundamentals indicate that Democrats are in better shape to take the Senate more than the polling does.
When I run the aforementioned fundamentals model to project every race and then simulate the national Senate election, the median result is that Democrats pick up a net gain of 2 seats. That's exactly the number they need to take to take control of the Senate.
Democrats are doing so well for three reasons. First, the national environment (as indicated by the national generic ballot) leans their way. Second, they have quality candidates running (either incumbent senators or former or current US representatives or governors). Third, Democrats are raising a ton of money from individuals. It's not just that money helps you buy ads, but it tends to flow to winning candidates (i.e., is a leading indicator).
The polling, however, shows something different than the fundamentals. When I run the simulations based solely on the polling, the median result is that Democrats pick up a net gain of no seats. That means Republicans will continue to hold a 51 to 49 seat majority in the Senate.
Pretty much all the incumbent Democratic senators in red states, such as Indiana's Joe Donnelly, Missouri's Claire McCaskill and North Dakota's Heidi Heitkamp, are doing worse in the polling than the fundamentals suggest that they should. None of these candidates are massive underdogs in the public polling. In fact, only Heitkamp is trailing and only by a small margin.
Yet, the simulations find that these incumbent Democrats lose a higher percentage of the time than the fundamentals because the polling is tight in all the races.
So is the polling or the fundamentals correct? FiveThirtyEight's Nate Silver has found that when the polls and fundamentals differ, the polls end up being closer to the truth on average. He also found, however, that the polling does tend to move closer to the fundamentals on average.
This shift towards the fundamentals may be exactly what is happening in Florida with Nelson.
If it begins to happen in other states, Democrats may have a better shot at winning control of the Senate than the polls would currently lead you to believe.