Inspiration
A Columbia University study found that nearly 25% of all trading volume on Polymarket is wash trading. We wanted to make this manipulation visible to everyday users.
What it does
We built on top of Polymarket's public API to score markets 0–100 on how likely their volume is fake, and profile individual wallets for insider-like behavior.
How we built it
Next.js app with four detection algorithms (self-trading, circular trading cycles, matched orders, volume anomalies) and Cytoscape.js network graphs to visualize suspicious wallet clusters.
Challenges we ran into
Polymarket's API only shows one side of each trade, so we had to infer counterparties by matching timestamps, sizes, and opposite sides.
Accomplishments that we're proud of
Our circular trading detector builds a directed wallet graph and finds cycles via DFS, surfacing coordinated wash trading rings invisible from individual trades.
What we learned
Wash trading detection is fundamentally a graph problem, which makes it interesting to solve.
What's next for PolyGuard
Persistent storage to track score trends over time, and on-chain wallet age analysis to catch fresh wallets created for manipulation.
Built With
- next.js
- polymarket
- supabase
- tailwind
- typescript
- vercel
Log in or sign up for Devpost to join the conversation.