Polymarket and Kalshi Turn Election Night Into a Margin Call for Your Personality

Tonight, America will discover who controls Congress and which guy in the group chat confused having a podcast subscription with running a quantitative trading firm.

November 3 offers a remarkable consumer upgrade: you can now watch your preferred candidates lose while your checking account experiences the same electoral map. For decades, election night merely ruined Thanksgiving. Polymarket and Kalshi have helped give it the power to ruin your November budget.

I do not want to be a moral snob about this. Adults can spend money on entertainment. People buy stadium beers, decorative swords, and subscriptions to men explaining why everyone else is a sheep. Putting twenty bucks on a race is hardly civilization’s final emergency.

But the earnest bullshit surrounding election betting deserves to be dragged behind a campaign bus. A wager does not become an intellectual achievement because the app uses percentages. Sometimes you are just a guy losing $2,000 on a Senate race in a state you could not locate without zooming out.

The billion-dollar number is real. Your imaginary loss estimate is not.

Reuters reported on Election Day that Kalshi had reported $1.33 billion in trading volume across midterm bets as of Monday. It also cited the Anti-Corruption Data Collective’s roughly $145 million figure for Polymarket congressional markets. Those are differently scoped figures, but they establish the scale: this is well beyond a few political obsessives passing a hat.

They do not establish that $1.48 billion is about to be lost.

Trading volume measures activity. Positions can change hands repeatedly. A contract’s maximum payout, the cash paid for it, the amount still exposed, and a trader’s eventual net loss are different numbers. Adding the loudest figures together and labeling the result “money Americans will lose tonight” would be financial reporting performed with a leaf blower.

A defensible total would require positions, purchase prices, sales, outcomes, and fees, with traders’ offsetting positions accounted for. These headline volume figures do not provide that. Nor does every election contract settle tonight.

The defensible conclusion is still ugly: across this enormous trading activity, people holding losing contracts through ordinary binary settlement will lose their entire purchase cost, plus applicable fees. Others will win. Some will have hedged. Nobody needs to invent a national bankruptcy counter to explain why this can get expensive.

A 95-cent favorite can still eat your entire $950

Polymarket’s own explanation describes the basic machinery: opposing outcome shares are backed by a dollar, and the winning outcome pays a dollar per share. The losing outcome becomes worthless. You can also sell before resolution at an available market price.

Here is the arithmetic for a hypothetical purchase of 1,000 ordinary binary contracts held to resolution, excluding fees. These are illustrations, not live quotes or forecasts:

That last row is where “basically free money” goes to have its organs harvested. At the same contract quantity and price, one 95-cent favorite losing wipes out the gross profit from nineteen winners. Fees worsen the arithmetic.

Meanwhile, a cheap long shot is cheap per contract. Buy enough contracts and your tiny contrarian masterstroke becomes a substantial invoice for disagreeing with reality.

For scale, 10,000 hypothetical bettors each losing $1,000 would mean $10 million in gross losses among those bettors. That is an illustration, not an estimate of tonight’s losing accounts. The important point is how little personal drama it takes to build a very large number: ordinary people, ordinary stakes, repeated thousands of times.

You do not need one spectacular whale to produce a financially miserable evening. You can assemble it out of dads.

The market can be smart while you are being an idiot

The strongest defense of prediction markets is also a terrible sales pitch for the casual punter: they can aggregate information effectively.

An October 20 research preprint by Harry Crane and Koleman Strumpf, using anonymized Kalshi data from the 2024 election, reports that prices could be accurate despite an unrepresentative trading population. It also finds that traders making just one trade put 73% of their volume on Harris, while more active traders accounted for most volume and showed less association between their location and their positions. These are historical findings from a preprint, not a measurement of tonight’s losses.

They are nevertheless an excellent antidote to the idea that signing up makes you part of the smart money by osmosis.

A market can correctly favor the eventual winner while plenty of participants lose. A well-calibrated 75% forecast should still fail about one time in four across comparable events. And buying a correctly priced contract does not create an advantage. If the true chance is 75% and you pay 75 cents for a dollar-or-zero outcome, your expected gross profit is zero. Costs move it below zero.

You need an advantage over the price you actually pay. Having a preferred outcome is available free to every idiot with a television.

We previously covered tools for comparing Polymarket and Kalshi. Better tools can make information easier to inspect. They cannot turn “my brother-in-law says turnout looks huge” into proprietary research.

The wisdom of crowds has no contractual obligation to make every member of the crowd solvent. Some members are there to be the tuition.

Congratulations, your civic engagement has transaction costs

The platforms are exchanges, so the simple story that every losing stake goes straight into the CEO’s pocket is wrong. In the basic matched contract, one side’s trading gain comes from the other side’s loss. Before fees, and setting aside external rewards or hedges, redistributing the money does not create investment returns for everyone.

The fees make the consumer proposition less charitable.

Under Kalshi’s published general fee formula, a 1,000-contract purchase at 50 cents with the default taker multiplier of one produces a $17.50 trading fee. That turns the $500 purchase into $517.50 paid and the winning profit into $482.50. Specific markets and intermediaries can have different charges; this is an illustration of the general schedule.

Polymarket’s international platform fee documentation lists a 0.04 taker-fee parameter for politics, applied to quantity times price times one minus price. At 1,000 shares and 50 cents, that formula yields $10 before applicable rebates. Its documentation distinguishes politics from fee-free geopolitics and describes maker and taker rebate programs. Do not confuse that parameter with a flat 4% charge on the amount staked, or assume every Polymarket venue and contract has identical terms.

Even the fees need explanatory journalism. Naturally, the purchase button fits on a phone.

There is nothing inherently illegitimate about charging for an exchange. But the customer who keeps jumping between positions needs to beat trading costs as well as other traders. If your entire edge consists of typing “the polls are cooked,” the fee schedule has brought a calculator to a feelings fight.

Election night is already stressful. Have you tried adding a sell button?

Tonight’s particularly nasty temptation is to turn every update into an instruction. A county reports. A pundit changes tone. Your position drops. Suddenly you are executing a sophisticated new strategy called Please Make The Red Number Stop.

Selling early can limit a loss. It can also lock one in before the outcome changes. An available exit price depends on the orders other people are willing to trade; the last displayed price is not a promise that your entire position can leave at it.

Nor should a television race call be mistaken for an automatic payout. Kalshi’s settlement guidance explains that payment follows confirmation and finalization under the market’s rules, with delays possible while it waits for required data. “Election night” and “money back in your balance” are separate appointments.

This is the same consumer transformation we examined in our deep dive on betting apps and sports culture: watching becomes wagering, and the event becomes a sequence of opportunities to transact. Our coverage of Schwab’s move toward yes-or-no market products tracked the same interface getting a more respectable address.

Now apply that impulse to an activity people already struggle to discuss without threatening to leave the family group chat. Wonderful. We have finally found a way to make political polarization involve more payment processing.

You are allowed to gamble. We are allowed to call it gambling.

I am not asking anyone to clutch pearls. I am asking the industry to stop pinning a little economist costume on the wager and demanding a standing ovation.

Prediction markets can produce useful information. Skilled traders can profit. Someone can place a modest recreational bet and have a perfectly pleasant evening. None of that changes the fate of a losing contract held to settlement, or turns the next deposit into evidence of civic sophistication.

The people worth being mean to are the salesmen of effortless certainty, the dashboard prophets, and the newly minted election savants who believe the market has somehow overlooked the polling average they found on the first page of Google.

The person who loses money deserves an honest account of what happened. They bought a risky position, paid a price, and got an outcome. The app’s success at forecasting Congress does not reimburse them.

America already lets you be loudly wrong about politics for free.

Polymarket and Kalshi have made that a premium experience.