I Analyzed 19,505 Polymarket Trades and Found 253 Suspicious Insider Trading Patterns
A comprehensive data analysis of insider trading on Polymarket — including coordinated wallet clusters, sybil networks, pre-resolution loading, and 100% win-rate wallets. Here's everything we found.
TL;DR
We built an algorithm that monitors Polymarket trades in real-time and flags suspicious insider trading patterns. After analyzing 19,505 trades across 2,652 markets and 5,276 wallets, we detected 253 insider trading alerts — including 50 critical-severity and 80 high-severity findings.
This isn't theoretical. These are real trades, real wallets, and real patterns happening right now.
Why We Did This
The headlines keep coming:
We wanted to know: are these isolated incidents, or is insider trading on Polymarket systemic?
The answer, based on the data: it's systemic.
The Numbers
| Metric | Value |
|---|---|
| Total trades analyzed | 19,505 |
| Markets monitored | 2,652 |
| Unique wallets tracked | 5,276 |
| Total alerts detected | 253 |
| Critical severity | 50 |
| High severity | 80 |
| Medium severity | 80 |
| Suspicious wallets flagged | 146 |
Alert Breakdown by Type
| Detection Signal | Alerts | Description |
|---|---|---|
| Pre-resolution trading | 51 | Large positions taken hours before outcomes are known |
| Wallet cluster coordination | 38 | Multiple wallets trading in lockstep |
| Profitable wallet anomalies | 37 | Statistically impossible win rates |
| Flash positions | 30 | Rapid large positions opened and closed |
| Sybil patterns | 29 | Same entity operating multiple wallets |
| Volume spikes | 24 | Abnormal volume concentration before events |
| Timing anomalies | 19 | Suspicious off-hours or burst trading |
| Wash trading | 18 | Wallets trading against themselves |
| Contrarian conviction | 7 | Large positions against market consensus that win |
Finding #1: Coordinated Wallet Clusters
The most damning finding. We detected wallet pairs trading across up to 20 shared markets — always in the same direction, with near-instant timing.
The Worst Offender
Our top critical alert: two wallets (one identified as "0x2b784d9d..." and pseudonymized "Unwieldy-Counsel") traded 20 markets together within 5-minute windows, with:
When two "independent" wallets trade 20 different markets with 100% directional agreement and 1-second timing, that's not two people thinking alike. That's one person operating two wallets.
Additional Cluster Activity
We also detected a second cluster: "Frail-Possible" and "Unwieldy-Counsel" trading 16 shared markets together with 100% same-direction trades and 2-second average gaps.
The "Unwieldy-Counsel" wallet appears in both clusters, suggesting a 3+ wallet operation controlled by a single entity.
Finding #2: Sybil Networks (Same Entity, Multiple Wallets)
We detected 29 sybil pattern alerts — cases where wallet behavior is so identical that they're almost certainly the same person.
What "100% Market Overlap" Looks Like
| Wallet Pair | Shared Markets | Overlap | Buy Ratio Diff | Time Between First Trades |
|---|---|---|---|---|
| Fat-Keyboard & Adolescent-Congressman | 6 of 6 | 100% | 0.0% | 0 hours |
| 0x14393c5e... & 0xf2ebb6fb... | 8 of 8 | 100% | 0.0% | 1 hour |
| Coarse-Catastrophe & Interesting-Flatboat | 5 of 5 | 100% | 0.0% | 0 hours |
| Orderly-Entity & Alarming-Fang | 5 of 5 | 100% | 0.0% | 1 hour |
| Alarming-Fang & 0xb8da74e6... | 6 of 6 | 100% | 0.0% | 0 hours |
When two wallets trade the exact same markets with 0.0% difference in buy ratios and their first trades are 0 hours apart, there's no reasonable explanation other than: this is the same person.
Notice that "Alarming-Fang" appears in two pairs — suggesting a 3-wallet sybil cluster operating together.
Finding #3: Pre-Resolution Position Loading
This is the classic insider trading signal: taking large positions when you already know the outcome.
Critical Example
Wallet "Leading-Countryside" took a $3,712 BUY position on "Over" in the Nets vs. Pistons O/U 215.5 market — 11 hours before resolution. This position represented 13.6% of the market's total liquidity.
More Pre-Resolution Alerts
The pattern is consistent: large, directional positions taken hours before outcomes on markets where the person clearly has information the market doesn't.
Finding #4: Impossible Win Rates
Multiple wallets have 100% win rates — but exclusively on low-liquidity markets where information advantages are easier to exploit.
Nobody is that good. A 100% hit rate across multiple markets is the hallmark of someone trading on foreknowledge, not analysis.
These wallets specifically target low-liquidity markets because:
Finding #5: Timing Anomalies
We detected wallets with 100% off-hours activity (1-7AM UTC, weekends) — patterns suggesting time-zone-advantaged information access.
Wallet "Bogus-Fix": 4 trades during off-hours worth $2,174, all in burst patterns. When someone only trades during hours that are convenient for one specific timezone and does so in bursts, it suggests they're acting on information they receive during their waking hours — not conducting research.
Our Methodology
4-Signal Composite Detection
The algorithm runs 4 independent detection signals:
1. Pre-Event Trading (35% weight) — Detects positions taken before significant price moves or market resolution. Uses statistical deviation from baseline volume to flag anomalous timing.
2. Win Rate Anomaly (30% weight) — Identifies wallets with statistically improbable win rates, weighted by market liquidity and trade volume to filter out lucky small traders.
3. Volume Spike Participation (20% weight) — Flags wallets that disproportionately participate in abnormal volume windows, especially when directionally correct.
4. Cluster Coordination (15% weight) — Graph analysis identifies wallets with correlated timing, same-direction trades, and shared market participation.
Validation
We backtested the algorithm against 13 known insider trading wallets from 5 publicly documented cases (ZachXBT/Axiom ring, Israel-Iran military trades, KPMG earnings manipulation), plus a control group of 50 clean wallets.
| Metric | Result |
|---|---|
| True Positive Rate | 92.3% |
| False Positive Rate | 0.0% |
| Overall Accuracy | 98.4% |
| Precision | 100% |
| F1 Score | 0.96 |
The algorithm correctly flagged 12 of 13 known insiders while producing zero false positives against the clean wallet control group.
What This Means for Polymarket Traders
If you trade on Polymarket, you need to know:
What Can Be Done
For Traders
Check the Insidex Dashboard before placing bets. See which markets are flagged and which wallets are suspicious.
For Polymarket
Implement cross-wallet position aggregation, volume quality metrics, and real-time cluster detection. The data is there — it just needs to be acted on.
For Regulators
Rep. Torres' "Public Integrity in Financial Prediction Markets Act" is a start. But prediction markets need integrity tools built into their infrastructure, not just regulations applied from outside.
The Dashboard
All of this data is live and free to browse at [insidex.nanocorp.app/dashboard](https://insidex.nanocorp.app/dashboard).
You can see:
Deep Dives
*If you found this analysis useful, share it. The more people who know about insider trading on prediction markets, the harder it gets for insiders to operate in the shadows.*
*Have questions about the methodology or want to discuss specific findings? Reach out or join the conversation on Reddit.*