Kalshi charges the confident and pays the patient

There was a slide in a client deck in 2016 that put the addressable market for a medical logistics product at $14bn by 2020, and every person in the review room knew the number was wrong. The analyst who built it knew, because he had multiplied three cells by a fourth and called the result a model. I knew, because I had told him which cells to multiply. The director knew, because he had asked for the number two days earlier and had not once asked how it was made. The figure went into the deck, the deck went to the steering committee, and the committee made a hiring decision on the strength of it. At no point did being wrong cost anyone in that room a single thing. The forecast was free, which is another way of saying it was worth exactly what it cost.

Kalshi is the opposite institution, and I mean that with more precision than the phrase usually carries. It is an exchange designated by the Commodity Futures Trading Commission in November 2020 and open to the public since July 2021, on which a claim about the future takes the form of a contract paying a dollar if the claim comes true. A forecast on Kalshi has a price, the price is set by people willing to be wrong for money, and the person making the claim can be made to hold it himself. If you believe an event has a 70 percent chance, you can buy that belief at 70 cents, and if the event does not happen your cents are gone. The deck was free. This costs. Everything worth saying about the difference lives in what the difference does to the people on each side of it.

The fee is small, and the fee is the finding

The taker fee is 0.07 multiplied by price multiplied by one minus price, so it peaks at 1.75 cents on a contract priced at 50 cents and shrinks toward nothing at the extremes, which is the schedule Kalshi's own help pages published and the fee aggregators reproduce.

Small numbers, and easy to dismiss, until someone counts them across an actual book of trades. Constantin Burgi, Wanying Deng and Karl Whelan did exactly that in a CEPR discussion paper published in September 2025, using more than 300,000 Kalshi contracts, and their findings have a shape the marketing does not mention. Cheap contracts lose money once the fee is counted; the longshot bias that horse-race bettors have carried for a century shows up here too, and the people who supply liquidity, the makers, out-earn the people who cross the spread, the takers. The confident pay the patient. I want to be careful with the venue, because this is a discussion paper rather than a peer-reviewed article, and the authors' names are on it while no sponsor's is, which is more transparency than most productivity research manages and still not the same as settled science.

My own position here is the one the paper predicts. I funded an account out of my runway, which makes me a taker by construction, and the takers are the ones who pay. An essay explaining the house odds is written by a man who has been paying them, and I'd rather publish that than pretend I opened the account for research purposes only. Both things are true: the account is small, and it is still the reason this essay exists.

The exchange earns on volume, and volume can be manufactured

That's the interesting part, because the fee schedule rewards the exchange for one thing only, trading volume, and volume responds to subsidies as readily as to conviction. For most of its life Kalshi ran a Volume Incentive Program that refunded fees to high-volume traders, and on September 28, 2026 it filed with the CFTC to terminate the program effective no earlier than October 13, scrapping a schedule that had been set to run into October 2027. The filing arrived amid allegations of wash trading, claims the company denies, and after a market-maker class action filed in November 2025 alleging the platform let sophisticated traders profit against its own customers, which the company also disputes. Consider the incentive. The affiliate, Kalshi Trading, sits on the maker side of the same book where its owner's customers are takers, and the company says the affiliate is not profitable, a claim nobody outside can audit.

The money around the exchange has moved faster than the forecasting on it. A $1bn Series F at a $22bn valuation closed on May 7, 2026, led by Coatue, with reports of later talks near $40bn that remain unconfirmed. Sports contracts arrived in January 2025 and crypto perpetuals followed between April and June 2026, and a CFTC request to list stock and ETF perpetuals is pending rather than live, so I'm leaving it alone. Meanwhile the states are arriving with lawsuits shaped like court decisions. New York's attorney general sued in July. The circuits disagree with one another: the Third Circuit sided with Kalshi in April, the Ninth Circuit sided with Nevada in August, and the Sixth Circuit let Ohio and Tennessee enforce their gambling laws in September. The legal map is now a patchwork, and a patchwork is a tax on the calmest part of the product.

The priced claim is the moral argument, and it is strong

The case for Kalshi deserves better than the marketing gives it, so here it is at strength. A committee forecast has no cost, which means it has no discipline, and a pundit's forecast has no cost either. The only known cure for costless conviction is making the conviction purchasable, because a man who must buy his own 70 percent claim at 70 cents discovers in real time whether his confidence survives contact with his wallet. Money-backed markets have beaten expert panels on calibration for as long as the records exist; the oldest finding in the field is that a thin market with real money outperforms a room of people whose being wrong costs nothing. The drift toward sports and crypto does not refute any of that. A casino-adjacent order book can still be the most honest forecast in the building, the way a casino's roulette wheel is the most honest randomizer in the building, and the steering committee I sat in never once cleared that bar.

It depends, as always, on which product you're actually buying. If the forecasting core keeps the liquidity that the incentive program was paying for, the exchange is the institution its defenders describe. If the volume was mostly subsidized and mostly sports, the forecast is a loss leader wearing the institution's clothes. The variable worth watching is what happens to the event books now that the subsidies are gone. The number in my old deck was free, and it moved a hiring decision. Kalshi's numbers cost their claimants money every day, which is the only reason anyone has to believe them.