Nate Silver recently expressed his puzzlement over Donald Trump’s reelection chances in betting markets improving from about 30 percent to 50 percent in recent months:

Silver subsequently pointed out that the new probability is a better reflection of Trump’s 50-50 chances this far out. What puzzles Nate is why the betting odds improved for Trump over the past quarter.

My hunch is that the uptick comes from the rash of stories showing fundamental election models giving Trump excellent odds for reelection. For example, back in March, Politico’s Ben White and Steven Shepard interviewed a bunch of these forecasters and concluded, “if the election were held today, he’d likely ride to a second term in a huge landslide, according to multiple economic models with strong track records of picking presidential winners and losses.” And before you, my dear reader, scoff at such models, do not forget that they predicted 2016 pretty darn well.

So, does that mean the odds of Trump’s reelection chances will continue to go up? Honestly, prediction markets are too thin and poorly incentivized to make an impression on me anymore. But consider today’s Spoiler Alerts my own gut check compared to this kind of estimate. (I plan on updating this once a quarter to provide a guide of my own thinking about 2020.)

I went into this exercise agreeing with Silver that Trump’s chances were about 50-50. The more I think about it, however, the slightly more bearish I get on Trump’s chances. So right now I’d say there’s more like a 45 percent chance of Trump winning reelection.

Why? Three reasons:

1) The economy will not get any better for Trump and probably will get worse. Those fundamental models are based primarily on economic data. White and Shepard wrote that the “White House remains confident that the GOP tax cut will support growth of 3 percent both this year and next, keeping job and wage gains strong.” Except that, oops, actual economic growth has tapered off and revised figures show it was not at 3 percent last year. Which means that all those models suggesting great odds for Trump will have to be revised downward.

Furthermore, the numbers that actually matter for reelection are what the 2020 economy looks like, because the election is not today but in November of next year. And the odds are quite good that the economy next year will be worse than now. With the tax cut and the budget deal, Trump has already shot all his bolts to stimulate the economy. What remains is an anti-growth mélange that imposes sanctions on the U.S. economy. If Trump were to end the trade wars, than would provide a boost, but if anything he’s trending in the more bellicose direction.

Furthermore, the Federal Reserve has started signaling that it won’t continue to match trade war escalations with interest rate cuts. St. Louis Fed President James Bullard, a relative dove, said in a speech this week that “I don’t think it is realistic for the Fed to respond to each threat and counter threat in a tit-for-tat trade war.”

To be clear, there are still some good economic numbers for Trump, such as the low unemployment rate. Still, contra Ramesh Ponnuru, I think this is less of a strength than would have been expected in the spring. And this has been Trump’s strongest card. The weaker it looks, the worse his chances for reelection.

2) Trump cannot count on Democratic apathy in 2020. A big reason Trump squeaked by in 2016 was that constituencies that traditionally vote Democratic were not quite as enthusiastic about Hillary Clinton as they had been about Barack Obama. For example, more than 66 percent of eligible African American voters cast ballots in 2012; in 2016 that number fell to 59.6 percent. Most of those who cast their ballot voted for Clinton, but turnout was clearly a problem.

Will African Americans be as apathetic in 2020 as in 2016? Unlikely. Trump himself has helped to ensure this by launching racist tirade after racist tirade. Over the weekend, the New York Times’s Reid Epstein and Jonathan Martin reported on this issue in terms that should petrify the Trump 2020 campaign:

Turnout figures show many stayed home in 2016, an election that marked the first decline in black participation rates in two decades. Increasing black turnout by just a few percentage points in urban areas of states like Michigan, Wisconsin and Pennsylvania could thwart Mr. Trump’s re-election.
And there are already signs that Mr. Trump’s conduct, which has been reminiscent of a 2016 campaign filled with racist tropes, is likely to ensure that outcome.
Longtime black Democratic leaders say the only time they can recall black voters being so engaged in presidential politics was when they had the chance to elect, and then to re-elect, Barack Obama.
“My dental hygienist talked with me about the election for 40 minutes the other day,” Shirley Franklin, a former Atlanta mayor, recalled with wonder, adding: “Some have preferences but a lot don’t. They just say, ‘I want to vote for whoever is going to beat Trump.’ That’s the predominant feeling.”

If Trump loses Michigan, Pennsylvania and Wisconsin in 2020, then the ballgame would be pretty much over. It is not as though there are states that Clinton won but might tilt toward Trump in 2020.

3) Democrats in 2020 are not Republicans in 2016. One of the more amusing elements of post-debate punditry has been to see right-wing commentators desperately trying to pump up Democratic presidential candidates who might roil the party, such as Marianne Williamson or Tulsi Gabbard. After each debate, a whole raft of commentators tried to make these candidates happen by highlighting Google searches or whatnot. It’s not easy to dismiss the possibility, as four years ago a fringe candidate managed to win the GOP nomination.

HuffPost’s Ariel Edwards-Levy throws some cold water on this prospect, however, by looking at how voters responded to all the candidates after the second debate. They showed that Elizabeth Warren, Pete Buttigieg and Cory Booker did really well; Gabbard and Williamson less so.

As for actual polling, Williamson remains close to zero and probably will not be on the debate dais in September. Indeed, for all the claims that Joe Biden’s support would crumble, it has held quite steady so far. And a big chunk of that support rests on the premise that Biden is the Democrat most likely to defeat Trump next year — a premise that the polling data to date bears out.

So, to sum up: Trump’s economy looks weaker than it did three months ago, and the trend line is worrisome for his reelection campaign. Democrats seem both energized and pragmatic. Neither trend bodes well for the 45th president.