Vista Market
Prediction markets for long tail events
Created on 1st March 2025
•
Vista Market
Prediction markets for long tail events
The problem Vista Market solves
Prediction markets don't really work for long tail events for these reasons:
- providing passive liquidity is hard
- creating a market is hard
- resolution is often slow and manual
Vista market solves these problems by
- creating a market with a new design, friendly for LP providers
- allowing natural language market creation
- agent based oracles
User Interaction and Data Flow
Users can create markets by prompting on farcaster, tagging @asset-trading-ch channel.
An example of market would be:
- Will the price of retardio coin be more than 1$ in five minutes from now?
At that point, our ai enabled market creation flow creates a market with - The description given by the user along with a certain starting probability (50% if not specified otherwise)
- Some collateral provided by the market creator in any token (e.g. 100$)
After the market creation, the trading process starts.
Users can deposit collateral in the hook and buy YES and NO tokens.
Each buy or sell will reflect in a different composition of YES/NO tokens in the pool, in turn altering the probability and payoff.
This is a link that shows how the math of the market works:
https://docs.google.com/spreadsheets/d/1sej9rpGZzfE3vJSv-dP2ifjCagU3cMp27CrEwJgjbEA/edit?gid=0#gid=0
Finally, markets can be solved by agents, that can check market condition and resolve them, enabling users to then claim tokens. We started with price based market, but we could extend these markets about anything.
The project architecture and development process
Market creation:
- an ai based development flow checks for mentions of a certain farcaster account and triggers market creation. Market data is stored in walrus.
Trading
- the AMM: Our system’s CPMM with Uniswap hooks offers continuous liquidity and a transparent, mathematically defined pricing mechanism. It depends on sufficient trading activity and arbitrage to shift prices as new information comes in, but it doesn’t force a price jump at resolution. Standard outcome token markets like Polymarket often use pari‐mutuel or order-book mechanisms, where pricing and payout adjustments are driven more by discrete order flow and aggregated bets rather than a continuous invariant.
Resolution
- the oracle is an agent that accesses price information and calls smart contracts functions based on those.
Product Integrations
- uniswap v4 hooks: creating the prediciton market
- coinbase agentkit for the ai oracle
- walrus to store data
Key differentiators and uniqueness of the project
-This is the only prediction market design using dex liquidity pools, which allows a variety of new markets to be supported.
-Using ai based oracles allow for faster resolution
-Creating markets with natural language to create virality on social media
Trade-offs and shortcuts while building
- the farcaster creation flow is not deployed on a server, so we would need to manually run a script to create markets
- the hook has been tested estensively, but the code is not really optimized
Additional Features
this project was built during the buidlathon. New features we will support:
Minimize LVR by withdrawing liquidity as market end approaches
Use a custom curve to mitigate loss as the market is ending
More info at market creation (e.g. volume P&L chart based on market set up)
Add new feeds and support more subjective markets
Make inference verifiable. Create a dispute resolution mechanism for edge cases.
Tracks Applied (9)
Multi-chain Attestations
Optimism
Walrus Tusked Champion
Sui | Walrus
Build an AI-powered app on Base
Base
Build a Farcaster MiniApp on Base
Base
Most Innovative Use of AgentKit
Coinbase Developer Platform
Consumer Crypto Prize
Coinbase Developer Platform
Best Developer Feedback
Coinbase Developer Platform
DEFI, NFTS + GAMING
Uniswap V4 on Unichain Innovation
Uniswap