What is functionSPACE?
functionSPACE is two things at once:
One where beliefs are expressed as probability curves over a numerical range, not as yes/no contracts.
So you can ship a prediction surface inside any product without running a venue, hosting liquidity, or operating a resolver.
In a regular prediction market you pick a side: "Yes, Bitcoin closes above $100k by year-end." Your only choice is direction. functionSPACE asks a different question: "Where will BTC actually close?" You answer with a curve — a shape over a numerical range that captures both your point estimate and how confident you are. The market sums everyone's curves into one consensus distribution. When the outcome resolves, your payout depends on how close your curve was to where the answer actually landed.
Every market in the competition is a simulated market. There is no real money at stake. functionSPACE generates simulated trades against the markets so they feel live — the consensus moves, prices respond, payouts compute — but you and your users are interacting with dummy markets seeded with paper liquidity. Build as if it were real, because the mechanics are real. The collateral and resolution are not.
Why curves: the market is a function, not an order book
The deepest shift in functionSPACE: the market itself is a function over the outcome space. Not a book of yes/no orders sitting at price points. A continuous mathematical object that everyone's beliefs additively shape.
This has three consequences.
functionSPACE markets are not "distribution-only" markets. They accept any shape a user wants to express — and the obvious ones are still here:
Every one of these expressions composes onto the same market function and contributes to the same pool of liquidity. The binary and range views that current prediction markets force you into are still natural — they're just one option among many, and they share liquidity with everyone else.
A binary venue covering "BTC year-end price" needs a separate market for every $5k bucket, each with its own thin book. functionSPACE puts all that capital into one shared pool covering the whole range, because there's only one function to update.
You don't just get a single price. You get the full distribution — including how confident the crowd is, not just where it thinks the answer lies. Your app can read any slice of it: a probability for any range, the mean, the variance, the consensus mode.
Try it: shape a belief
The two surfaces
Every functionSPACE market is designed as two linked surfaces sharing the same ID, range, and precision.
Where beliefs are traded as curves and the consensus forms. This is the surface you build against.
The resolution layer. A capital-backed, adversarial mechanism that converges on the final outcome. It replaces the role an external oracle plays in most prediction markets — instead of trusting a third-party data feed to declare the truth, capital-backed assertions and slashing converge on a numerical value the protocol settles against.
The Reality Market is not yet live, and the markets themselves are dummy markets. Resolution isn't part of what you're building against. You only need to think about the Probability Market.
For traders
Submit a belief, get claims, watch the consensus move, exit or settle. Five steps, fully collateralized.
Pick a market, express your view in whatever shape fits — a single point, a binary above/below, a range, a confidence-weighted distribution, or any combination — and decide how much to put in.
Claims are your weight in the consensus. You get more claims per dollar when the pool is small, or when your view diverges from current consensus. Early and contrarian = cheaper influence.
Your curve is added to the running market total. The live consensus is everyone's curves, weighted by claims.
Sell your position back at its current mark-to-market value. If consensus moved toward your view, you profit. If it moved away, you lose.
Your payout = your claims × how close your curve was to the actual outcome (scored via CRPS, a strictly proper scoring rule). Inaccurate positions get nothing. The pool always settles to exactly its size — no external capital needed.
For builders
functionSPACE is a primitive. The protocol provides the mechanism. You build the product on top.
The protocol owns: market state, belief math, claim minting, mark-to-market pricing, settlement scoring, payout distribution. You don't implement any of this.
You own: what your app looks like, how users input beliefs, what data you surface, who can trade, how you monetise, what you embed it inside.
Three integration depths
Trade panels, charts, position tables already work. Pick a theme. Ship.
Use hooks like useMarket and useConsensus, render whatever you want.
Vanilla TypeScript, framework-agnostic, full control over every interaction.
You don't have to pick one — mix and match. Full reference at docs.functionspace.dev.
What you can build
The protocol is domain-agnostic. The constraint isn't the mechanism — it's where you think prediction belongs.
Sports stats, election forecasting, earnings probabilities, weather, esports brackets.
Belief curve painters, mean+confidence sliders, presets, mobile-first UX, terminal-style power tools.
Dashboards over consensus distributions, calibration leaderboards, market discovery feeds.
Bots that read public data and submit calibrated beliefs autonomously.
Consensus distributions as oracle feeds, structured products on top of belief curves.
Drop a market into a forum, chat app, newsletter, or Discord.
Intro video
A short walkthrough of the primitive from @functionspaceHQ.
FAQ
No. Every competition market is simulated, with paper liquidity and simulated trades generated by functionSPACE so the consensus moves and feels live. No real collateral is at stake. Build it as if it were real — the mechanics are real — but nothing in the competition resolves to a real outcome with real payouts.
No. The competition uses dummy markets, so settlement isn't part of what you're building against. In production, resolution is the job of the Reality Market.
Market creation is not currently supported via the SDK. If your build needs a specific set of markets, reach out and we'll spin them up for you.
Yes. Use the trade panel only. Use just the consensus API as a data feed. Build a read-only analytics tool with no trading at all. You're not forced into the full stack.
No. The SDK lets you translate simple inputs — a slider, a yes/no toggle, a confidence meter, a range selector — into a belief under the hood. Users only see your UI.
There are no LPs. Every trader funds the pool by depositing collateral when they take a position. The pool grows as people use the market, and the market deepens with participation.
Yes. A position can be exited at any time and is repriced against the current market state. If the consensus moved toward your view, you exit with more than you put in. If it moved against you, you exit with less.
Influence costs money. The more the market state has already moved, the more capital you need to push it further. Late, low-conviction trades barely move consensus. The market gets harder to shift as it grows.
You decide. The SDK supports gas abstraction — sponsored, relayed, or paid in stablecoin — so you can build experiences that feel fully Web2: email login, no wallet UI, no chain awareness. Power users can connect their own wallet if you want to support that path.
Yes. In production, the SDK supports developer fees of 0–5% on buys, sells, or both. You keep what you collect. Fees aren't enforced or paid out in the competition demo.
Chain selection is ongoing. functionSPACE is designed to be multi-chain compatible, and the production deployment target will be confirmed ahead of mainnet launch. The competition runs on simulated infrastructure, so chain choice doesn't affect what you build.
Where to go next
Pick the path that matches what you came here to do.