CrowdIntel4 crypto wallets: 98% win rate across 875 resolved bets
4 Polymarket wallets, all funded from a single on-chain address, placed 875 resolved bets on crypto markets and won 98% of them — a result you would expect less than 1 in 10,000 of the time by chance. They moved $57K in volume for -$5K in net profit.
The evidence
Across 875 resolved bets the cluster won 98% outright. Adjusting for sample size, the Bayesian win rate is 97% at 98% confidence, and the result sits 27.0 standard deviations above a 50/50 baseline. A coincidence of this size would occur less than 1 in 10,000 of the time.
How the wallets are linked
These wallets share the funder 0xc590…c02b, but that address behaves like shared infrastructure (an exchange, relayer, or the Polymarket proxy factory) rather than a single coordinator, so the "shared funding" link is NOT treated as evidence of coordination here.
Cluster highlights
Funding map
The wallets
The cluster is made up of 4 wallets, each linked on-chain to the same funder. Individual records below.
| Wallet | Bets | Win rate | CC | Streak | PnL | |
|---|---|---|---|---|---|---|
KU kubicanosl 0x5add…d0df | 120 | 100% | — | — | -$1.3K | |
MM mmmm44445551 0xd62f…f49b | 17 | 100% | — | — | -$4.4K | |
MA maxdone 0x8621…fe25 | 52 | 87% | — | — | +$223 | |
LO London9 0x3014…3f81 | 686 | 98% | — | — | +$917 |
Markets they targeted
The wallets concentrated their bets in a handful of crypto markets, ranked here by the cluster's volume.
Methodology & limitations
How we detect these clusters, what the numbers mean, and what we do not claim.
How we detect it
Every Polymarket trade is read directly from the Polygon blockchain. We trace the USDC that funded each trading wallet back to its source address. Wallets sharing a single funder are grouped, their funder is screened against exchange / relayer / proxy-factory infrastructure, and only genuinely coordinated funders are reported as clusters. Their market outcomes are then aggregated from on-chain resolutions.
What the numbers mean
- Win rate — resolved bets won ÷ total resolved.
- Bayesian win rate — win rate shrunk toward 50% to correct for small samples; closer to raw as the sample grows.
- Confidence — how much the sample size supports the estimate.
- Z-score — standard deviations above a coin-flip baseline.
- By chance (p-value) — how often a result this strong would happen by luck.
Funding linkage
A “verified” funding link means the wallets received USDC from the same address and that address is not a known exchange, relayer, or the Polymarket proxy factory. We currently trace direct (one-hop) funding; multi-hop and same-batch timing analysis are in progress.
What we do not claim
We do not assert the identity of any wallet owner, their intent, or that any law was broken. Shared funding plus an improbable win record is evidence worth examining — not a verdict. Figures are computed from on-chain data and update as markets resolve.