Landing Page
Open Markets
Trading Interface
Real Time Activites
Ai agent

Marketloo

start trading now -> https://market-loo.vercel.app/

Inspiration 🚀

Our team loves competing and saw hackathons were missing something fun. Every hackathon has those moments where everyone's predicting winners and debating what teams will build - so we turned those discussions into a prediction market where hackers can bet and climb the leaderboard. Think "Polymarket meets MLH" but with fake internet points called POO (Prediction Outcome Options).

What it does 🎯

MarketLoo is a prediction market platform where users trade virtual shares on real-world outcomes. Users can create markets, take positions on binary outcomes (Yes/No), and compete for profits. The platform features real-time price updates, a dynamic order book, and automated market making to ensure constant liquidity. Each user starts with 10,000 POO tokens to trade with.

We created AI trading bots with quirky personalities that dive into market data and trade alongside you! From the cautious "Warren Buffett Bot" to the YOLO-trading "WSB Bot", each has its own trading style. Race against them to the top of the leaderboard and see if you can outperform our algorithmic friends! 🤖📈

How we built it 🛠️

Built with Next.js 14 and Supabase, MarketLoo uses a serverless architecture for scalability. The frontend employs Tailwind CSS for styling and React for dynamic components. The backend features a custom order book system with automated market making, and a dual-resolution time series database for efficient price history tracking. Real-time updates are handled through Supabase's PostgreSQL subscriptions, while authentication and user management leverage Supabase Auth. The platform uses server-side rendering and API routes for optimal performance. ⚡

Challenges we ran into 😅

*Time-Series Data Magic *📊 One of our biggest headaches was figuring out how to handle all that sweet market data without making our database cry. We got clever with it and built a two-table system that's like having both a short-term and long-term memory:

recent_price_history: For the fresh, hot data (5-minute snapshots)
archived_price_history: For the old stuff, but organized neatly

Playing Market Maker 🎪 Making a market feel alive without real users is like throwing a party before anyone shows up. So we built this cool market-making system that:

Keeps the party going with automatic trades
Uses some fancy math to make prices move naturally
Makes sure there's always someone to trade with

Our secret sauce? This formula that makes everything work smoothly:

priceImpact = baseSpread + (random * liquidityFactor) + (volumeImpact * log10(orderSize))

Accomplishments that we're proud of 🏆

We actually managed to plan things out (shocking, we know!) and it worked! Instead of our usual "code first, think later" approach, we spent time designing our systems properly. The result? A trading platform that doesn't fall apart when you look at it funny! Plus, we made everything look pretty slick - no sketchy UI here!

What we learned 🧠

Turns out, breaking big scary problems into smaller, less scary problems actually works! Who knew? 🤷‍♀️ We also learned that we could build way more complex stuff than we thought possible - guess those Stack Overflow sessions paid off!

What's next for Marketloo 🚀

We've already got some RAG stuff and automated trading with AI agentsworking, but we're planning to add smart market analysis based on DevPost history to support traders. 📚

🏆 Prize Tracks:

Finance and AI (Tempo Labs YC23) Most Technically Impressive Beginner tracks (20 winners) Best Personal Project (CodeBuff) STATSIG GRAND PRIZE Colega AI prize Magic loops

All the boring data pipeline stuff is ready - we're just waiting to make our dreams come true! 🌟