Ever since Republicans’ Obamacare repeal efforts stalled and then died in the D.C. machine, investors worry that stocks have “overpriced the ability of the administration” to enact new policies, as a Goldman Sachs strategist recently put it. Simmering congressional conflicts will certainly not make it easier.

One place to look for clear-eyed political prognostication? You could do worse than online betting markets, which give Trump a solid chance of enacting at least some key proposals. Here’s a sampling of the latest polling, as of April 20 from PredictIt:

• Will there be an individual tax cut by the end of 2017? 60% no
• Will the individual mandate be repealed by the end of 2017? 66% no
• Will Donald Trump still be the president by the end of 2017? 82% yes

Currently if you take wagerers’ word for it, by the end of 2017 Americans will pay less in taxes, the GOP is marginally likely to get a deal for corporate tax cuts, Obamacare repeal will go nowhere, and Donald Trump is very likely to still be the president. There’s not an infrastructure market on the platform (yet), but PredictIt offers a 20% chance that debt levels will rise between $500 and $750 billion—meaning betters expect least some of Trump’s legislation will be passed. Right now, with no further action, the Congressional Budget Office expects debt levels to fall over the next year.

Take all of this with a handful of salt. The accuracy of betting markets have been questioned in the recent years after they failed to predict both Brexit and Trump’s own election.

But Microsoft Research economist David Rothschild says their predictive powers aren’t well suited to large events with lots of available polling data. They’re actually most useful in smaller, complex climates, like Trump’s D.C. “Where markets really have a lot of strength is where there’s idiosyncratic data that’s hard to write a model for,” Rothschild says. On questions like “whether or not a certain piece of legislation will pass, they seem to be pretty well calibrated,” he says.

Academic studies have repeatedly suggested that crowd-sourced predictions have merit. At PredictIt, which works with dozens of universities to hone the prediction model concept, all betters are limited to $850 and each question is limited to 5,000 people. The average number of shares traded per day is 687,760, but can go higher. On April 2nd, for example, the site had 747,885 shares traded on Trump-related markets alone. “The whole concept is wisdom of the crowds with a big enough crowds that you can guess something as obscure as the weather,” says Brandi Travis CMO for Aristotle International, the political software company that services PredictIt.

In the Trump era, prediction markets will likely be gauged against the actual stock market. A series of stocks considered to be a barometer for whether Trump will pass a border tax seems to be predicting that he won’t. And the market as a whole seems to expect that he’ll be able to goose the economy in a significant way.

If you’re an optimist, you might bet that prediction markets herald some economic progress. Otherwise? They’re one more way for people to lose money if the Trump Bump fizzles.

A version of this article appears in the May 1, 2017 issue of Fortune with the headline “Wagering on Policy.”