On November 3, US midterm election voting day. In less than 24 hours, the Democratic Party successively won the House of Representatives and the Senate, sending Democrats to the Senate in states like Texas and Kansas for the first time in decades.

The renowned polling agency Polymarket was also busy; today's Polymarket is far from what it used to be, handling events like elections with ease.

On Polymarket, almost every key election has a corresponding trading market: who controls the House, how many seats the Republicans can hold, which swing states will turn blue. For example, the probability of "Republicans hold the Senate" was only 25% on the eve of the election; this number not only represents the market-priced probability itself but also means that if the Republicans hold the Senate, those who bet "yes" in advance will get back about 4 times their principal.

Two years ago, the most famous election trade in the history of prediction markets was born here. According to CBS's "60 Minutes," a French trader self-funded a polling agency to conduct "neighbor polls," based on which he judged that Trump was undervalued by the market, then heavily bet on Trump's victory and earned over $80 million.

This time, there was no second French whale. But the experiences of three accounts illustrate better than any profit leaderboard how the money flowed in this election round.

2 bets earned over $1 million

Before hitting this midterm election, the most notable trade of account 0xa2cd was a crushing defeat.

On June 14, the probability of "US and Iran reach a permanent peace agreement before June 15" reached 76%. Two days earlier, Iranian state media had reported that the two sides had drafted a 14-page memorandum, and the market widely expected the agreement to be announced soon.

But 0xa2cd chose the other side, investing $800,000 betting on "will not reach"; hours later, Trump announced that the US and Iran had reached an agreement, and 0xa2cd's $800,000 was entirely wiped out.

More than four months later, he once again stood opposite the majority.

On November 2, the day before the vote, 0xa2cd began buying a seemingly quite extreme outcome: the Republicans lose all Senate elections in seven states: Alaska, Iowa, Kansas, Maine, Michigan, Ohio, and Texas.

0xa2cd invested $610,000 when the probability of this event of Republicans being swept was around 30%.

On voting day, he also bought "Republicans will not win Texas" at about 74% and pocketed $106,000.

As major mainstream media raced to report Democrats winning key swing state after swing state, the probability of the seven-state combination gradually rose. As of publication, six of the seven states have basically been settled, and the probability of this event has risen to 94%. 0xa2cd's paper profit on this position is nearly $1 million. Combined, he made about $1.1 million in this round of midterm elections.

But the money is not truly in hand yet. Among the seven states, only Alaska has no result yet. Republican incumbent Senator Sullivan and Democrat Peltola are still locked in a tight race. Alaska uses ranked-choice voting, and the final count may take some time. If Sullivan holds the seat, this position will go directly to zero.

Made $7 million on the World Cup, lost $1.58 million predicting this round of elections

Standing opposite 0xa2cd is an account far richer than his.

BreakTheBank has accumulated nearly $8 million in profit on Polymarket, of which about $7.1 million came from this summer's World Cup. After the World Cup ended, he moved to the NBA and NFL, making $1.09 million on a single Grizzlies vs. Jazz game.

However, for him, politics is clearly not as easy to predict as sports. In the same seven-state market, BreakTheBank believed Republicans would not lose all seven states and put in about $1.8 million. On this single trade alone, he is currently sitting on an unrealized loss of about $1 million; including his other midterm election positions, his paper loss this round is about $2.05 million.

The money made predicting Trump's victory 2 years ago has now given back $130,000

If the books for 0xa2cd and BreakTheBank are not yet closed, sweetspea's account is basically clear.

This account was registered in October 2024, catching the final weeks of that year's presidential election. At the time, Polymarket already gave Trump more than a 60% chance of winning, while mainstream polls still believed the two sides were neck and neck.

sweetspea sided with the market, placing heavy bets that Trump would win the U.S. presidency; these bets earned him approximately $370,000 in profit.

Two years later, on October 28, the midterm elections became one of the most discussed topics across the internet, and sweetspea made another heavy bet, confident that Republicans would hold the Senate: when the probability of "Republicans hold the Senate, lose the House" was 25%, he put in about $80,000.

Several other trades also revolved around "Republicans win": for the Ohio Senate election, he bet on Republicans; for the Ohio gubernatorial election, he also bet on Republicans.

All 9 midterm election positions lost money, totaling $135,000, more than one-third of his 2024 election profits.

About 40% of the entire election sector made money, but profits only flowed to large holders

As of November 4, 1,286 midterm election-related events on Polymarket had cumulative trading volume of $475 million. Of the more than 80,000 accounts participating in these events, about 30,000 made money.

But the number of people making money and the amount of money made are two different things.

All profitable accounts together earned about $21.83 million. Among them, the top 1% by profit, about 300 accounts, took 76.4%; the top 5% (about 1,500 accounts) took 94.4%. The top ten on the profit leaderboard earned a combined total of about $4.21 million, nearly one-fifth of the entire profit pool.

The remaining nearly 29,000 profitable accounts received only about $1.22 million in total, averaging about $42 per account.

The losing side is even more concentrated. All losing accounts together lost about $21.24 million, with the top 1% by loss, about 470 accounts, bearing 87.6% of it.

In other words, what was really trading in this election was a small group of large holders. The probabilities cited by the media and used as polling references were largely set by their capital.

This also raises the core question about the narrative that "prediction markets are truth machines": can probabilities priced by a minority still be trusted?

At least this time, the answer leans toward yes.

In the market for "Democrats control both the Senate and House," on September 17, the probability stood above 60% for the first time, and for over a month afterward it mostly fluctuated between 60% and 66%; it rose to around 70% in the week before the vote, and reached 75% on voting day.

Before the first ballot was counted, the market had already priced a Democratic sweep of both chambers as a high-probability event. sweetspea's 25% trade bought precisely the outcome the market deemed only a one-in-four possibility.

70% is not 100%. It does not mean the outcome is predetermined, nor does it mean those betting the other side were necessarily wrong—in that 30% scenario, they could have been the winners. But it at least provides a more concrete reference than slogans and polls: under the constraint of real money, what the market believes is more likely to happen.

At this moment, that reference is still operating in Alaska. The probability of Peltola winning is 94%, and the swing of over $6 million in profit and loss between 0xa2cd and BreakTheBank rests on that number.

Whether the market is right this time will have to wait until Alaska finishes counting the last ballot.

Welcome to join the official BlockBeats community:

Telegram Subscription Group: https://t.me/theblockbeats

Telegram Discussion Group: https://t.me/BlockBeats_App

Official Twitter Account: https://twitter.com/BlockBeatsAsia