$3.5 million is a small cheque by fintech standards. What makes the one Furnace just closed interesting is where it went: not into a flashy consumer app, but into the plumbing underneath event markets, funded by backers its website lists as North Island Ventures, Reciprocal Ventures, Reverie and the asset manager VanEck, with angels including DTCC and Public.com, as reported by casino.org. When institutional money starts paying for reference data, the category has stopped being a novelty.
So here are prediction markets explained without the hype, because if you already bet on sport or play casino games, the mechanics are close enough to feel familiar and different enough to cost you money if you assume they are the same thing.
What a prediction market actually is
A prediction market is an exchange where people trade contracts tied to the outcome of a real-world event. Each contract typically pays out a fixed amount, often $1, if the stated outcome happens, and nothing if it does not. The current trading price, say $0.62, is the market’s collective estimate that the event will occur: roughly 62%.
That is the whole idea. You are not accepting a price from a bookmaker. You are buying a yes/no contract from another trader who disagrees with you, at a price the two of you negotiate through an order book. If the market thinks an event is near certain, the yes contract trades near $1 and the no contract trades near $0, and the upside on each side shrinks accordingly.
Resolution is the part people skip, and it is the part that bites. Every market has a resolution source written into its terms: an official result, a government data release, a named authority. The contract settles against that source, not against what you thought the question meant. Two venues can run what looks like the same market on the same event and settle it differently because one wrote a tighter deadline or a different data provider into the terms.
Why $3.5 million went into infrastructure, not an app
Furnace emerged from stealth with a $3.5 million seed round, positioning itself as what it called the reference layer for prediction markets. Its stated audience is institutional: hedge funds, prime brokers, proprietary trading desks, plus developers building products on top of existing venues. Covered venues include the US and international versions of Polymarket, with another major operator already supported and more being added, delivered through integration or APIs.
What Furnace reportedly builds
The problem it targets is contract fragmentation. Take interest rates. One venue might list a market on the Federal Reserve cutting rates before the end of 2026, another on a cut within a specific month, another on a cut outside a stated window. To a human these look related. To a risk system they are three unrelated rows of data.
Furnace’s approach, in its own words, is to give each economic exposure a persistent canonical identity and map venue-specific instruments underneath it, while keeping each venue instrument distinct with its own terms and resolution details. Translated: it tells a trading desk which contracts are genuinely the same bet and which only look alike.
That is not a product retail punters will ever touch. It matters because it is the kind of unglamorous fintech gambling investment that only makes sense if you expect serious volume to arrive. Nobody builds identifier systems for a market that stays small.
Walking one market from open to settlement
The lifecycle of an event trading market runs in five steps:
- A venue opens a market on a question with a defined outcome, deadline and resolution source.
- Traders post bids and offers. The price settles where buyers and sellers meet, between $0 and $1.
- That price doubles as an implied probability. $0.30 means the market collectively prices the outcome at about 30%.
- News, data and money move the price. You can sell your position before resolution, taking a profit or a loss at the current price rather than waiting for the result.
- The event resolves. Winning contracts pay the full amount, losing contracts expire worthless.
Step four is the real departure from a betting slip. A settled wager at a sportsbook is locked until the event finishes. A contract can be traded out at any point the market has liquidity, which means your exit price, not just the final result, determines what you make or lose.
Who sets the price: the market, the bookmaker, or the RNG
The three formats answer four questions very differently.
| Prediction market | Sportsbook (fixed odds) | Casino game | |
|---|---|---|---|
| Who sets the price | Traders, via an order book | The bookmaker, adjusted for liability | Nobody; paytable is fixed against RNG outcomes |
| Built-in edge | No structural edge in the price itself; venue takes fees | Yes, the vig built into the odds | Yes, the house edge, fixed and known |
| Who funds the payout | The trader on the other side of your contract | The operator’s book | The operator, from long-run margin |
| What decides the result | A named real-world resolution source | The official result of the event | A random number generator |
Put numbers on it. A sportsbook offering 1.91 on both sides of a coin-flip market is implying about 52.36% for each (1 ÷ 1.91), which adds to roughly 104.7%. That 4.7% overround is the bookmaker’s margin, present before the ball is kicked. European roulette carries a 2.7% house edge on every spin, and most slots run a 94% to 97% RTP, meaning a 3% to 6% edge to the operator. These are not estimates. They are the design.
Prediction market pricing has no such built-in margin baked into the price. This is the line you will see repeated in every pitch deck for sports prediction markets, and it is true as far as it goes. Here is the term that qualifies it: no structural edge is not the same as a favourable edge. You are trading against counterparties who may be better informed, faster, or using exactly the kind of institutional tooling Furnace sells. Removing the house does not remove the person on the other side of your contract.
Gambling, derivative, or something in between
Legally, this is unsettled, and anyone telling you otherwise is selling something. Event contracts sit awkwardly between two bodies of law: financial regulation, which treats binary payoffs on real-world outcomes as derivatives, and gambling regulation, which licenses operators jurisdiction by jurisdiction. Which framework applies to a given platform depends on where it is based, where the user sits, and how the specific contract is written.
What is observable is that the venues themselves are adopting gambling-style protections. Polymarket has introduced deposit limits, self-exclusion tools and other safeguards, per casino.org’s reporting. Operators do not add self-exclusion to a pure derivatives product. Treat any claim about legality in your own jurisdiction as something to verify with your local regulator rather than something to infer from a platform’s marketing, and check current guidance before depositing. [Editor: attach current regulator guidance for the relevant jurisdiction here.]
What thin liquidity and fees quietly do to your returns
Four practical points before you fund an account on any of the prediction market platforms:
- Liquidity thins fast away from headline markets. A contract on a major election trades tightly; an obscure question may show a wide gap between the best bid and the best offer, and that gap is a cost you pay on entry and again on exit.
- Fees are not always visible in the price. Some venues charge on trades, some on winnings, some on withdrawals. Model them before you assume a 55%-priced contract on a 60% event is profitable.
- Counterparty and platform risk is real. Your funds sit with the venue. Custody, solvency and withdrawal processing are all part of your risk, separate from whether your forecast is right.
- Being right is not enough. If a contract already trades at $0.85 and the event happens, you made 15 cents on 85 at risk. Returns come from being right at a price the market has mispriced, which is a harder job than picking winners.
And the obvious point that needs saying plainly: this is risk-taking with money you can lose in full. The absence of a house edge does not make it an income strategy. Set a limit before you start, treat losses as the cost of participating, and use the deposit caps and cool-off tools the platform provides if your spend starts drifting.
A category worth watching, not a sure thing
The honest reading of Furnace’s raise is modest: a small seed round into reference data for a market that institutions are still circling, not pouring into. The bolder reading, which the backer list supports, is that identifier infrastructure is what a market builds immediately before volume arrives. Both are defensible. I lean towards the second, because the fragmentation problem Furnace describes only becomes worth paying to solve once desks are trading across multiple venues at once.
What that does not tell you is whether these venues will be regulated as exchanges, as sportsbooks, or as something new, or whether retail traders will find them any kinder than the markets they already know. Betting innovation has a habit of arriving before the rulebook. The thing to watch over the next year is not the next funding announcement. It is which regulator claims jurisdiction first, and what they decide these contracts actually are.
