Only On Fuel is a research series featuring innovative projects and founders in the Fuel ecosystem.
For Swaraj and the team at Griffy, reinventing prediction markets wasn't just about creating another betting platform—it was about solving fundamental technical challenges that have long plagued the industry. "If you look at all the order book style prediction markets that exist today, including Polymarket, they often do off-chain order matching with blockchain settlements," explains Swaraj, Griffy's founder. "We feel there's still a chance of transactions getting censored or orders being unfairly matched when the system is off-chain."
This insight led the eight-person team based in Pune, India to build something different: a fully on-chain prediction market that leverages Fuel's unique technical capabilities to deliver a more transparent and efficient system. Since launching on Solana in July 2023 and migrating to Fuel on mainnet day, Griffy has facilitated over 21,000 prediction polls, with their largest single bet reaching an impressive $30,000.
Reimagining prediction markets with on-chain order books
Traditional prediction markets face a critical challenge: they typically rely on Automated Market Makers (AMMs) that require significant capital to bootstrap each new market. "In an AMM setup, you would need someone to put in a couple thousand dollars for each event," Swaraj explains. "And since one party always loses in a prediction market, you're practically guaranteed to lose that bootstrapped capital."
Griffy's solution is an order book-style prediction market that eliminates the need for this initial capital. Users can create simple yes/no predictions about current events—from sports matches to political outcomes. The platform operates on a precise matching system where a 'Yes' order at price x is matched with a 'No' order at price y, provided x + y ≥ 100 (where x and y are in cents). When orders are matched according to these conditions, winners receive their profits minus a small 1% fee.
Predicates and UTXOs on Fuel
The decision to build on Fuel came down to its unique technical capabilities, particularly its predicate system and UTXO model. At its core, a predicate functions like a programmable vault, where the "password" is a set of conditions that must be met for the funds to be spent. As Swaraj explains, "Think of predicates as a function: if supplied input equals ABC, then return true." When the conditions are met, specialized "solvers" can execute transactions by satisfying these predicate conditions, earning a small fee for this service.
This predicate system, combined with Fuel's UTXO model, offered Griffy a solution to a fundamental challenge in building decentralized order books. "Implementing order books on EVM-based chains presents significant challenges in terms of computational costs and state growth. The increasing state size makes it prohibitively expensive to maintain order books at scale," says Swaraj.
Griffy leverages these capabilities in an innovative way. While traditional smart contracts maintain state on-chain, Griffy uses predicates as stateless programs that define spending conditions for each bet. When a user places a bet, they create a predicate address that holds their funds. The predicate includes specific conditions for spending, such as the odds they're willing to accept and the outcome they're betting on.
After an event concludes, an oracle provides a signature confirming the outcome. Users holding tokens for the winning outcome can then redeem their USDT through what Griffy calls the "settlement predicate." This system provides several advantages over traditional smart contract implementations, including no need for users to hold native tokens for gas, automatic order matching without centralized intermediaries, and efficient state management through the UTXO model."
Features and future development
Griffy's platform currently offers binary yes/no predictions on current events, an order book-style matching system. The automated settlement process and 1% fee structure on winning positions only have helped drive user adoption and engagement across thousands of prediction markets.
Looking ahead, the Griffy team has ambitious plans for expanding their platform's capabilities. One of the most exciting developments involves integrating zero knowledge proofs into Griffy's oracle system. "Once the ZK opcode capability goes live on the Fuel VM, we could have a ZK-TLS integration where UMA still exists on, say, Polygon, and UMA results can be used to settle things on Fuel," Swaraj explains. This integration would allow Griffy to automate result verification while maintaining decentralization and connect with established oracle networks on other chains.
A new community model: prediction markets for everyday users
What sets Griffy apart isn't just its technical architecture—it's how the platform democratizes prediction markets for everyday users. Unlike traditional betting platforms or prediction markets that require significant capital or technical expertise, Griffy allows anyone to participate with any amount they're comfortable with. Users aren't betting against a house or a complex AMM system—they're matching with other users who hold opposing views, creating a more balanced and fair ecosystem.
This peer-to-peer approach, combined with Fuel's efficient predicate system, means that whether someone is wagering $30 or $30,000, they have the same opportunity to participate and earn rewards. While not every prediction will be correct, the platform's continuous stream of new polls—spanning from sports outcomes to ecosystem developments—means users always have fresh opportunities to apply their knowledge and insights.
"What we're building isn't just about predictions—it's about creating an engaging way for people to participate in events they care about," explains Swaraj. As Griffy continues to evolve with Fuel's expanding capabilities, particularly the upcoming ZK features, it's positioning itself not just as another prediction market, but as a new model for how decentralized platforms can create more inclusive and efficient financial products.