With the first Trump–Biden debate now smoldering behind us, the betting markets have picked their winner. That also means they've decided on a loser: President Donald Trump.

Tuesday marked the first time the incumbent squared off face-to-face against his rival, Democratic nominee and former Vice President Joe Biden. The two septuagenarians bickered, insulted, belittled and at times screamed at one another at almost regular intervals, all in hopes of convincing the American public each is the best choice to control the world’s largest deployed nuclear arsenal.

And while conventional, real-time polling may have been unavailable during the roughly hour and a half Trump and Biden squabbled, bets were furiously being made over whom the markets believe will be the person running the U.S. for the next four years.

So why turn to the so-called predictions markets, where people bet on the outcomes of major events? The theory goes something like this: Those willing to risk their capital provide more accurate information about expectations than those who merely answer a survey. Money talks, as they say.

On PredictIt, a prediction market run by New Zealand’s Victoria University of Wellington, the volume of “shares” on the question “Who will win the 2020 U.S. presidential election?” saw its highest-ever day on Tuesday, coming in at about 708,500.

Bettors buy “shares” in a candidate. When a candidate wins, the payoff is $1 per share; everyone who bet on someone else goes home with nothing. Besides being centralized, PredictIt is also regulated in the United States.

Shares of “Trump” started off the debate trading at around 46 cents, roughly where they had been for much of the month. Early in the debate, as it quickly heated up, buyers began growing a little more optimistic on Trump’s prospects, with shares rising to as high as 48 cents. However, they closed the hour back at 46 cents after roughly 46,000 shares traded hands. Just seven hours before, fewer than 5,000 shares were traded.

As the night wore on and the debate took a turn for the weird, if not troubling, sellers came in where they could, taking prices down to as low as 42 cents, more or less where it was at press time.

Read more: Prediction Markets’ Time Has Come, but They Aren’t Ready for It

Meanwhile, Joe Biden’s shares remained more or less in the 59-cent range during the debate. However, some 39,700 shares were traded. Considering the higher price, that’s a couple of thousand dollars more in volume. Biden’s stock has since been on an upswing, as high as 63 cents as this article was published.

Given the nature of this election, there were a few trades for the nominees’ running mates.

Some 40,000 shares of Kamala Harris were traded between 3 cents and 5 cents during the debates. A buyer or several buyers, perhaps speculating the GOP would want to or need to change the top of their ticket, bought nearly 100,000 shares of Mike Pence at the for between 1 cent and 2 cents.

The American election, of course, isn’t a one-horse race. Rather, it is 51 separate elections for sets of electors from each state and the District of Columbia. And that’s where some of the more interesting price movements have been taking place.

On Wednesday, bettors changed their minds about Florida, moving the Sunshine State from Republican at 52 cents to Democrat at 51 cents. Like presidential shares, state races also pay $1 to the winner and nothing to the loser.

That compounds trouble on the electoral map for Trump, who is seeing traditional southern Republican states like North Carolina and Arizona come into play. Wagers also appear confident Wisconsin, Michigan and Pennsylvania – the three states crucial in Trump’s 2016 victory – will switch to the Democrats.

The moves now give the Democrats a likely 335 votes in the Electoral College, up from 306 several hours before. This puts PredictIt’s numbers close to Nate Silver’s FiveThirtyEight.com’s prediction that Biden will capture 332 electoral votes. Candidates require 270 electoral votes to win the presidency.

Bets on the Senate haven’t been kind to Majority Leader Mitch McConnell, either.

Gamblers now see the Democrats picking up enough seats to downgrade the longtime Kentucky Republican senator to Chuck Schumer’s (D-N.Y.) current office as Minority Leader, assuming Republicans in the upper chamber decide to keep him at their helm even if they lose control. According to PredictIt, Democrats will start the next session with 51 senators, picking up states including Arizona, North Carolina and Maine.

While PredictIt is a centralized market, more blockchain-friendly exchanges also have bad news for Team Trump.

To be sure, they are smaller in size than others but offer more tailored questions. And since this is an anything-goes election, such bespoke contracts may help those who need to hedge for some reason or else just want to take on an enticing bet. After all, whether Trump will concede the election should he lose remains a mystery.

For example, Augur, the most well-known decentralized prediction market, offers one contract on “Will Donald J. Trump win the 2020 U.S. Presidential election?” expiring Jan. 20, 2021. Another one – “Will Donald Trump win the 2020 U.S. Presidential election?” – expires Jan. 7, 2021. “Who will win the 2020 U.S. Presidential election?” expires on Dec. 7, 2020.

In the first two, a “yes” contract (that Trump wins) is at 49%. In the last one, Trump is at 40% to Biden’s 54%. Presumably, the 6% is an unknown factor which, in 2020, is entirely possible.

Polymarket has bettors pricing a Trump win at 46 cents though briefly Tuesday night, it was as high as 52 cents. Just $100,000 worth of contracts have traded on that contract since it began three months ago. PredIQt on the EOS blockchain gives the president nearly the same odds, trading at 45%. Omen, which prices its contracts in DAI, gives Biden a 58.8% chance of winning.

FTX, a market that doesn’t just trade crypto derivatives, also has a market on various election-related contracts. Traders there are also bearish on Trump.

Trump’s “futures contract” on FTX had a rapid sell-off just as the debates came to a close. Prices fell to 40 cents from 43.5 cents on volume of 57,000 contracts. Back in February, it traded as high as 65.2 cents before crashing to as low as 31.8 cents in the midst of the March 12 market crash. It rebounded to 50 cents until June, it began trending lower.

Biden’s price got a boost at about the same moment, albeit on smaller size. His contract on FTX rose to 60.4 cents from 55.5 cents on 10,000 contracts just as the debate commentators were shaking their heads on-air in disbelief at the spectacle they had just witnessed.

Of course, prices fluctuate every day.

Facts change and, as this is 2020, sometimes get invented, too.

Thus what’s true at the time of publication can change on a dime. It is now fewer than five weeks until Election Day. Buckle up!

Correction (Oct. 1, 01:00 UTC): Corrected status of U.S. nuclear arsenal — it is the largest deployed in the world, not largest overall.