My, how political times have changed here in Michigan.

A few weeks ago, Hillary Clinton seemed to enjoy a stranglehold on Michigan in the presidential race which, if her double-digit lead held in November, would probably cement the state’s status as a Blue State going forward for some time.

Since then, the race has tightened to a 38-35% lead – a statistical tie – for Clinton over Donald Trump.

Now, stats guru Nate Silver of FiveThirtyEight.com has listed Michigan as tied for the second-most important state in the entire general election.

Silver announced today that Michigan has become a tipping point in the Electoral College results, based on Five Thirty Eight’s latest forecasting models. Florida is the most pivotal, but it’s only the tipping-point state 16 percent of the time, based on various projections. After Florida, the crucial swing states of Pennsylvania, Michigan and Ohio each prove decisive in about 11 percent of Silver’s simulations.

To be clear, Silver says that Clinton’s chances of winning the presidency are 80% if she wins Michigan and only 9% if she loses Michigan. The math involved here, given the many possible variations of the Electoral College results, is such that Trump has a 91% percent chance of winning the general election if he carries Michigan and a 20% prospect of victory if he doesn’t.

michigan-no-2-pivotal Obviously, Michigan is much more of a must-win for Clinton than for Trump.

Meanwhile, Clinton’s overall likelihood of emerging victorious in November, according to Five Thirty Eight, is now down to 59%, a substantial drop from her standing in early August when Silver’s complex computer projections showed Clinton with an 80%-plus opportunity for a win.

Here’s Silver’s summary of the presidential race at this point in time:

Among the swing states, there are particular ones that Trump and Clinton need to avoid losing. If their opponent is ahead there, it’s an almost a sure sign to the candidates that they’re losing the election. For Trump, these are the Light-Red States (Ohio, Florida and Iowa); for Clinton, they are the Light-Blue States (Pennsylvania, Michigan, Minnesota and Virginia).

Nate Silver

Silver

A Clinton win in Florida would be a very bad sign for Trump. In 29 percent of our polls-only model’s simulations, Clinton wins the election while losing Florida; Trump wins just 6 percent of the time when losing the Sunshine State. It’s as close to a must-win for Trump as a battleground state can be.  The last time a Republican won the presidency without Florida was in 1924. Trump is just the slightest of favorites in Florida at the moment, with a 52 percent chance of winning the state — which gives you an idea of why he’s still an underdog in this election.

Other nearly must-wins for Trump include IowaNorth Carolina and Ohio. Trump holds small leads in all three states but has less than a 60 percent chance in North Carolina and Ohio. Now, it’s not as if Clinton doesn’t want to carry these states. She wins only about a third of the time when she doesn’t, but she doesn’t need them like Trump does. Pennsylvania, on the other hand, is close to a must-win for Clinton. She wins only 13 percent of the time when she loses that state. Trump wins a slightly higher 21 percent of the time without it.

Two other states that Clinton pretty much needs to win are Virginia and Wisconsin. Virginia lined up nearly perfectly with the national vote in 2012 but has continued to move left since then, and Clinton rarely wins (in simulations) without it. That might explain why Trump has spent a lot of money advertising in the state even as Clinton continues to lead there. Wisconsin is a different story. It has gone Democratic in every election since 1984, but Trump is hoping whites without a college degree help him turn the state red. He has not led in a single high-quality poll in the state but did close the gap in a recent Marquette University poll