faro
← Logbook31 October 2026 · Logbook · 1 of 3

Prediction markets in 2026: what they are, how they got here, and why most of the world says no

Logbook · 1 of 3. Written 31 October 2026; the numbers are as of that date and will age.

This is the first of three entries. This one is about the markets themselves. The second is about Faro: why we built it and what a week of real order-book data taught us. The third, written once our first experiment closes, is about what paper trading should reveal.

The idea in one paragraph

A prediction market lets people trade on the outcome of a question with a clear answer: Will the Fed cut rates in December? Who will win the Georgia Senate race? Each outcome is a contract that pays $1 if it happens and $0 if it doesn't. If the "Yes" contract trades at 62¢, the market as a whole is saying "about a 62 % chance". Anyone who thinks the real chance is higher buys; anyone who thinks it's lower sells. The price moves until the disagreement balances out. That is the whole trick: a price that doubles as a probability, set by people with money on it.

How it differs from the markets you know

Stock market Sports betting Prediction market
What you hold A share of a company A bet with a bookmaker A contract that pays $1 or $0
Who's on the other side Other investors The bookmaker Other traders (the venue just matches orders)
When it ends Never, in principle When the game ends When the question resolves, by rule
What the price means What the company is worth The bookmaker's odds, with a margin A probability
Can you exit early? Yes Rarely Yes, at the current price

Four things follow from that design, and they matter later:

  • It's zero-sum. Every dollar won is a dollar someone else lost, minus fees. There is no dividend or growth to share.
  • The rules decide everything. A contract is only as good as the wording of its resolution rule. Two venues can list "the same" question and settle it differently.
  • Prices should add up. If an election has five candidates, exactly one wins, so the five "Yes" prices should sum to about $1. When they don't, someone can in principle buy all five for less than the $1 they're guaranteed to get. That's arbitrage, and it's what Faro measures first.
  • Most markets are thin. A handful of headline questions get millions of dollars; most get very little. A price you can see is not always a price you can trade at in size.

A short history

Prediction markets are older than the internet hype suggests. The Iowa Electronic Markets have run small, university-licensed election markets since 1988. Intrade, based in Ireland, was the famous one in the 2000s and shut down in 2013 after US regulators sued it. PredictIt ran small capped markets in the US from 2014 under a regulatory "no-action" letter. All three were curiosities: interesting to economists, tiny in money.

Two companies changed that.

Kalshi (founded 2018 in New York by Tarek Mansour and Luana Lopes Lara) took the regulated road. In 2020 it became the first exchange the US derivatives regulator, the CFTC, approved specifically for event contracts. When it tried to list contracts on US elections, the CFTC said no (2023). Kalshi sued and won in 2024, just weeks before the presidential election. In early 2025 it added sports, and sports changed everything: today most of Kalshi's money is on games.

Polymarket (founded 2020 by Shayne Coplan) took the crypto road: a website, not an exchange, settling in a dollar-pegged stablecoin, open to anyone. In 2022 the CFTC fined it $1.4 million for offering unregistered contracts to Americans, and it blocked US users. It became famous anyway during the 2024 US election, when more than $2.6 billion was traded on the presidential race alone. In late 2025 it bought a small licensed US exchange, got the CFTC's approval, and returned to the US as a separate, regulated platform — Polymarket US — next to its global one.

Where they are now

The growth over the last year is hard to overstate:

  • Volume. Combined monthly trading on Kalshi and Polymarket went from under $5 billion in September 2025 to about $24 billion in April 2026 (Pew Research), then $45 billion in June and a record $50.6 billion in July, driven by the football World Cup. For scale: US legal sportsbooks handled about $14 billion a month in 2025.
  • Who leads. Kalshi overtook Polymarket in volume during 2025 and now does roughly three quarters of the combined total ($37.7 billion of $50.6 billion in July).
  • What people trade. Mostly sports. Since mid-2024, sports were about 80 % of Kalshi's volume and about 39 % of Polymarket's; politics was about a third of Polymarket's, and crypto most of the rest.
  • Money behind them. Kalshi raised $1 billion this week at a $40 billion valuation. Polymarket took up to $2 billion from Intercontinental Exchange, the owner of the New York Stock Exchange, in October 2025.

One detail from our own data: about 98 % of the markets listed on both venues are automated, short-lived contracts — 15-minute crypto windows, daily weather ladders, single games, multi-leg parlays — created and settled by the thousand every day. The "event" markets most people picture are a small minority of what's listed.

Why so restricted?

Almost everywhere outside the US, the law looks at a prediction market and sees gambling: you stake money on an uncertain outcome and win or lose it. Gambling needs a national licence, and these platforms don't have one, so regulators block them. Polymarket alone has been blocked or listed as unauthorised in Switzerland and France (late 2024), Poland, Singapore and Belgium (early 2025), Portugal, Brazil, Spain, Indonesia, South Korea and Denmark (2026), among others. Italy, where Faro is built, put Polymarket back on its list of unauthorised gambling sites in July 2026. A few places are moving the other way: Gibraltar is creating a dedicated regime, and the UK's financial regulator confirmed in October 2026 that it is talking to both Kalshi and Polymarket.

In the US the question is different: are these contracts financial derivatives (federal, CFTC) or sports bets (state law)? The platforms say federal; states say gambling, especially for sports. The courts disagree with each other. The Third Circuit sided with Kalshi in April 2026; the Sixth and Ninth Circuits sided with the states in a run of rulings since August, including Ohio and Tennessee on 25 September. On 1 October the Supreme Court declined to take the question up immediately, so the split stands. Meanwhile New York is seeking billions from Kalshi, Nevada and Minnesota have restricted access, and Baltimore became the first US city to sue both platforms (21 October).

Two other worries come up again and again:

  • Insider trading. If you know the outcome before others, a prediction market pays you for it. The CFTC issued a warning about trading on non-public information in February 2026.
  • Elections. Some regulators and politicians don't want money riding on votes, whatever the price says about them.

Why this matters for Faro

Three things in this picture shaped how we built Faro:

  1. Prices are information. Even where you can't trade, a market that thousands of people bet on is one of the best real-time probability estimates available. That makes prediction markets worth studying in their own right.
  2. The structure invites a specific question. Because prices should add up and outside forecasters should agree with markets about as often as markets are right, the gaps are measurable. Are they real, after fees and at real depth? How often? How big?
  3. Access is the constraint, not the idea. Most of the world can't trade these venues, and the rules are still being written. So Faro is a research tool first: it measures, and it tells you what the evidence says before anyone thinks about acting on it.

The next entry is about exactly that: what Faro is, how it works on one modest server, and what a full week of Polymarket order books — every change to every order book — showed when we looked closely.

Sources (all accessed 31 October 2026)