CrowdIntelDefinition
What is Funding cluster?
A group of Polymarket wallets that received their initial USDC from the same source address and often coordinate trades — a signature of Sybil behavior or organized insider activity.
In detail
Every Polymarket proxy wallet is funded by at least one address that sent it USDC. When many wallets trace back to the same funder and then bet the same direction on the same markets within tight timing windows, the cluster is probably operated by one entity. Funding clusters are one of the strongest signals for coordinated manipulation, Sybil attacks, or hidden insider activity, because the on-chain funding trail cannot be faked without additional gas expense.
How CrowdIntel measures it
CrowdIntel's USDC indexer ingests every inbound USDC transfer to Polymarket proxy wallets directly from Polygon, then groups wallets by shared funder. ClusterDetector combines funding topology with trade timing to score coordination. Investigations open when a cluster has ≥3 wallets, ≥20 combined bets, and statistically significant win rate (p-value ≤0.1) against the market-category baseline.
Frequently asked
Is every funding cluster suspicious?
No. The top funder on Polymarket (0xf70da9...) sent USDC to 601 proxy wallets — it's Polymarket's own infrastructure. CrowdIntel's cluster scorer scales penalties by cluster size and excludes platform infrastructure from insider classification.
What's the minimum cluster size that's actually coordinated?
2-3 wallets with shared funder + same-market bets within 60 minutes is a weak signal. ≥5 wallets with same-funder + same-direction bets + similar sizing in a 30-minute window is strong evidence of one operator.
Can clusters hide behind multiple funding hops?
Partially. CrowdIntel currently traces 1-hop funding; multi-hop obfuscation (fund A → fund B → fund wallets) defeats detection. This is a known gap and a roadmap item.
Related terms
- Polymarket insider
A Polymarket trader who bets with non-public information or against the consensus and wins at rates statistical models cannot explain by skill alone.
- Wallet clustering
The process of grouping blockchain wallets that are likely operated by the same entity, using shared funding sources, timing correlation, and behavior similarity.
- P-value (trading)
The probability that an observed win rate could occur by chance alone if the trader had no edge — used to separate genuine skill from lucky streaks.
- On-chain trading analysis
The practice of drawing trading insights directly from blockchain transaction data rather than exchange-reported metrics — enabling verifiable, un-gameable analytics.