The Problem with Binary Prediction Markets
Traditional prediction markets force users into rigid yes/no choices. When betting on Bitcoin reaching $100,000, participants with 70% conviction place identical bets as those with 99% certainty. This creates several problems:
- Loss of information — Market signal is diminished when nuance gets discarded at entry
- Unfair odds — Participants with different conviction levels end up in the same pool
- Limited expressiveness — Markets cannot capture the full belief spectrum
- Opaque settlement — Many platforms settle off-chain, requiring trust in centralized operators
A New Primitive: Percentage-Based Predictions
On Heavymath, every prediction is a number between 0–100 representing the participant's probability belief. An 80% prediction means "I believe there is an 80% chance this happens."
This structural shift captures far more information from its participants compared to traditional binary markets.
The Dual-Boundary Market Split
The core innovation is a market split algorithm that fairly divides participants into two sides based on percentage predictions and stake amounts.
When a market deadline passes, the algorithm identifies two boundaries:
- Negative boundary: predictions at or below form the negative side
- Positive boundary: predictions at or above form the positive side
- Middle zone: predictions between boundaries receive full refunds
The boundaries use a bidirectional capital-adequacy constraint ensuring each side has sufficient capital for worthwhile odds. Boundary stakes get trimmed proportionally if one side is overweight.
Desirable properties:
- Fairness — No forced poor odds
- Capital efficiency — Only necessary capital is locked
- Expressiveness — Full 0–100% range preserved
- Determinism — Pure math executed on-chain
How a Market Works: End to End
1. Market Creation
Dealers holding Dealer License NFTs create markets by specifying description, category, deadline, and optional oracle configuration. Dealers earn settlement fee shares, incentivizing quality curation.
2. Prediction Phase
Before the deadline, users select a percentage (0–100) and stake USDC. Predictions can be updated or withdrawn anytime before closing. No order books, counterparty matching, or liquidity pools required.
3. Market Lock
After the deadline, the split algorithm runs, determining boundaries and each participant's side. Middle-zone predictions are refunded; boundary predictions may be partially trimmed.
4. Resolution
Markets resolve to binary outcomes (positive/negative) via dealer action, on-chain oracles, or Chainlink Functions requests. Resolution method cannot change post-creation.
5. Settlement
Winners claim proportional shares from the distributable pool minus a settlement fee (default 1%), split between dealer and platform. All math happens on-chain.
The Dealer System: Decentralized Market Curation
Anyone can mint a Dealer License NFT to create markets within specific categories. This creates a curated ecosystem where market quality is maintained by the people who have the most domain expertise.
Dealers earn settlement fee shares from markets they create, aligning incentives with quality and participation.
The permission system uses category/subcategory hierarchies with wildcard support for fine-grained control.
Built-in Communication: Email Notifications via Signic
Heavymath integrates with Signic, a decentralized email platform. Every wallet automatically corresponds to a free email account.
Market events — locks, resolutions, refunds — trigger HTML notifications with relevant details: which side you're on, effective stake, win/loss status, and claim instructions.
This integration is built into the indexing infrastructure, triggered directly by on-chain events rather than a bolt-on service.
Sports: The First Vertical
While supporting any market topic, sports is the first deeply integrated vertical. The platform provides live data for ten sports: soccer, basketball, American football, baseball, hockey, rugby, Formula 1, MMA, handball, and volleyball.
Dealers create markets directly from upcoming fixtures with oracle-based resolution. Users browse leagues, teams, and games with full context — standings, head-to-head records, historical data — within the same interface.
Radical Transparency
Every aspect is verifiable on-chain:
- Market creation parameters, deadlines, and oracle configuration are public from inception
- Every prediction — percentage, amount, timestamp — is recorded on-chain
- The market split algorithm is deterministic and independently verifiable via smart contract
- Settlement fees, dealer shares, and payout calculations use fixed formulas with no discretionary components
There is no hidden order book, no market maker taking the other side, and no admin key that can change the rules after bets are placed.
Fee Model
Heavymath charges a settlement fee (default 1%) on the distributable pool — capital remaining after middle-zone refunds and boundary trimming — split between the dealer and platform.
No fees apply to placing predictions, updating them, or pre-deadline withdrawals. Users only pay when winning, with fees deducted from payouts.
Looking Ahead
The roadmap includes:
- Additional oracle integrations for broader market type automation
- Mobile applications with Wallet Connect integration
- Governance transitions for community control of platform parameters
- Expanded verticals beyond sports into elections, economic indicators, and cultural events
Try Heavymath
The platform is live on testnet at dev.heavymath.com, allowing exploration and prediction placement without real funds.
Conclusion
Prediction markets work best when they capture the full range of human belief. Binary markets throw away information. Heavymath preserves it.
The platform combines percentage-based odds with deterministic market splits, decentralized settlement, dealer incentives, email notifications, and sports integration to create more expressive and transparent prediction markets.
The future of forecasting is not yes or no. It is a number between 0 and 100.