Skip to main content

Polymarket Insider & Fresh-Wallet Detection API

Detect suspicious trading on Polymarket using on-chain data from Polygon. This page shows how to flag potential insiders by combining fresh-wallet checks, USDC funding-source tracing, wallet clustering, large bets, and pre-resolution timing.

Bitquery runs its own blockchain nodes and indexes, decodes, and parses raw Polygon transactions. That gives you signals the off-chain Polymarket Gamma and Data APIs cannot provide, such as how old a wallet is, where its money came from, and which other wallets share the same funder. These on-chain signals are the foundation of every serious insider-detection and copy-trading tool.

API Key Required

To query or stream data outside the Bitquery IDE, you need an API access token. See How to generate Bitquery API token ➤.

How Insider Detection Works on Polymarket​

A suspected insider leaves a specific footprint on-chain:

  1. A brand-new wallet with little or no prior activity.
  2. Funded by a traceable USDC source, often shared with other suspicious wallets.
  3. A large bet in a low-volume or niche market.
  4. Entered shortly before resolution, on the winning side.
  5. Redeems the winnings and goes dormant.

No single signal is proof. You combine them into a score and flag wallets that cross a threshold. The queries below produce each signal.

Signal to Data Map​

SignalWhat you measureBitquery source
Fresh walletWallet age (first on-chain activity) vs first betEVM.Transfers (dataset: combined)
Funding sourceFirst USDC transfer into the walletEVM.Transfers (USDC.e)
Wallet cluster (Sybil)Other wallets funded by the same senderEVM.Transfers grouped by receiver
Large bet in niche marketBet USD vs market 24h volumePredictionTrades
Pre-resolution timingHours between trade and resolution, winning sidePredictionTrades + joinPredictionManagements
Won and dormantRedemption amount, then no activityPredictionSettlements + Transfers
Why dataset: combined

The wallet-history queries below use dataset: combined so they search the full chain history, not just the realtime window. This matters for wallet age and funding traces, where you need the earliest activity ever recorded.

Step 1: Catch Large Trades in Real Time​

Start by streaming large Polymarket trades. Each event gives you a buyer address to investigate. Change subscription to query for historical results.

Run in Bitquery IDE

subscription {
  EVM(network: matic) {
    PredictionTrades(
      where: {
        Trade: {
          OutcomeTrade: { CollateralAmountInUSD: { gt: "5000" } }
          Prediction: { Marketplace: { ProtocolName: { is: "polymarket" } } }
        }
      }
    ) {
      Block {
        Time
      }
      Trade {
        OutcomeTrade {
          Buyer
          Seller
          CollateralAmountInUSD
          Price
          IsOutcomeBuy
        }
        Prediction {
          Question {
            MarketId
            Title
          }
          Outcome {
            Index
            Label
          }
        }
      }
      Transaction {
        Hash
      }
    }
  }
}

Step 2: Is the Buyer a Fresh Wallet?​

Look up the buyer's earliest on-chain activity. If the wallet's first transfer is close to the time of its first big bet, it is a fresh wallet and scores high. Replace the address with the buyer from Step 1.

Run in Bitquery IDE

query FreshWalletCheck {
  EVM(network: matic, dataset: combined) {
    Transfers(
      where: {
        Transfer: { Sender: { is: "0xe8a2057abd53d285f7bea590b8f1dff1f04454c2" } }
      }
    ) {
      earliest: Block {
        Time(minimum: Block_Time)
      }
    }
  }
}

A recent earliest time means a young wallet. You can also add count to the selection to get the wallet's lifetime transfer count, another freshness signal.

Step 3: Trace the Funding Source​

Find where the wallet's money came from. The first inbound USDC transfer is usually the original funder, often a centralized exchange withdrawal or a parent wallet. This query uses bridged USDC.e on Polygon (0x2791bca1f2de4661ed88a30c99a7a9449aa84174).

Run in Bitquery IDE

query FundingSource {
  EVM(network: matic, dataset: combined) {
    Transfers(
      orderBy: { ascending: Block_Time }
      limit: { count: 5 }
      where: {
        Transfer: {
          Currency: {
            SmartContract: { is: "0x2791bca1f2de4661ed88a30c99a7a9449aa84174" }
          }
          Receiver: { is: "0xe8a2057abd53d285f7bea590b8f1dff1f04454c2" }
        }
      }
    ) {
      Block {
        Time
      }
      Transfer {
        Sender
        Amount
      }
      Transaction {
        Hash
      }
    }
  }
}

Step 4: Find the Wallet Cluster (Sybil Detection)​

Take the funder from Step 3 and list every other wallet it funded. Wallets sharing a funder are likely controlled by the same operator. A large cluster placing correlated bets is a strong signal.

Run in Bitquery IDE

query SiblingWallets {
  EVM(network: matic, dataset: combined) {
    Transfers(
      limitBy: { by: Transfer_Receiver, count: 1 }
      where: {
        Transfer: {
          Currency: {
            SmartContract: { is: "0x2791bca1f2de4661ed88a30c99a7a9449aa84174" }
          }
          Sender: { is: "0xe8a2057abd53d285f7bea590b8f1dff1f04454c2" }
        }
      }
    ) {
      Transfer {
        Receiver
      }
      received: sum(of: Transfer_Amount)
    }
  }
}

Step 5: Timing and Outcome​

The final signals come from the trade itself. Check how soon before resolution the wallet entered, and whether it was on the winning side.

  • Pre-resolution timing: join the wallet's trades to the market's Resolved event and compute the lead time. Use the joinPredictionManagements pattern from the Polymarket Sports API.
  • Won and dormant: confirm the payout with a Redemption query on PredictionSettlements, then re-run Step 2 to check the wallet went quiet afterward.
  • Realized profit and win rate: see Realized PnL & Win Rate for Polymarket Trader.

Scoring Model​

Combine the signals into a weighted score and flag wallets above a threshold. Tune the weights to your tolerance for false positives.

score = 0.30 * fresh_wallet
      + 0.25 * shared_funder_cluster
      + 0.20 * bet_pct_of_market_volume
      + 0.15 * entered_within_24h_of_resolution
      + 0.10 * won_and_went_dormant

flag if score >= threshold

This mirrors how tools like CrowdIntel and PolyTrack work. Bitquery provides the on-chain primitives they otherwise have to reverse-engineer.

Limitations​

Read this before acting on a flag.

  • These are heuristics, not proof. A fresh wallet or a shared funder is not evidence of wrongdoing. Exchange withdrawal addresses fund many unrelated users and create false clusters.
  • The "entered before the news broke" idea needs an off-chain news timestamp. On-chain you can only proxy it with the resolution time or a sharp odds move (see the OHLC query on the Sports API page).
  • Always verify the current collateral token and any neg-risk or Conditional Token Framework specifics in the IDE before relying on results.
NeedAPI
Live odds, trades, volume by sportPolymarket Sports API
Trader realized PnL & win rateRealized PnL & Win Rate for Polymarket Trader
User & wallet activityPolymarket Wallet & User Activity API
Trades, prices, whalesPolymarket API
Settlements & redemptionsPrediction Settlements API
Real-time: Kafka streamsKafka Streaming Concepts

Support​