x402watch

Wash Report

Open methodology, anonymized findings. Updated daily.

Updated 2026-05-02 17:28 UTC

Most x402 dashboards count transactions. We classify them.

Total wash overview

Active buyers (30d)
2,476
Real volume %
48.9%
after wash filter
Suspected wash
144
5.8%
Self-test detected
277
11.2%

Wash share — last 14 days

Daily share of transactions classified as suspected_wash and self_test.

Pattern types

Every active buyer (30-day window) carries one of eight labels, applied in priority order. Labels are stable signals — not judgments — and reproducible from public on-chain data.

Pattern distribution

Active buyers (30d) by label.

  • organic_user1,678 / 71.8%

    A real user with diverse, unpredictable browsing patterns.

    Variable inter-tx gaps · mixed services · irregular hours.

  • self_test277 / 11.9%

    Operator-side wallet calling its own service — not real demand.

    Same wallet hits its own service shortly after deploy; small bursts.

  • developer218 / 9.3%

    Heavy bot using ≤2 services with near-constant intervals.

    ≥80 tx on each of 1–2 services · uniform spacing · long horizon.

  • suspected_wash144 / 6.2%

    Cohort or vanity cluster gaming volume metrics.

    Many wallets · uniform price · coordinated start within minutes.

  • ai_agent20 / 0.9%

    Autonomous agent making varied API calls with adaptive timing.

    Multi-service · variable prompt patterns · retries on failure.

  • analytics_bot0 / 0.0%

    Read-only monitor or data-collection script.

    Predictable cron · GET-shaped traffic · narrow surface.

  • exchange_user0 / 0.0%

    Wallets sourced from CEX hot-wallet patterns.

    Multi-hop deposit traces · short bursts · withdrawal cadence.

  • verifier0 / 0.0%

    Validator/oracle node spot-checking outputs.

    Repeated identical queries · signature verification cadence.

Anonymized case studies

Real patterns observed in the network — service names and seller addresses redacted. Operators may recognize their own footprint; others should treat this as a methodology preview.

Sophisticated Sybil Farmconfidence 0.90

Service A — Sophisticated Sybil Farm

Pattern observed

  • ·All buyers paying exactly $0.02 (uniform amount: 97%)
  • ·All started within a 12-minute window (coordinated start: 88%)
  • ·Each making 78–79 transactions (tx count CV: 0.23)
  • ·Random wallet addresses (no vanity pattern)
Buyers in cluster
60
Wash classification
93.4%

Signals matched

uniform_amountcoordinated_startuniform_tx_count_cvcohort_size
Vanity Clusterconfidence 0.97

Service B — Vanity Cluster

Pattern observed

  • ·17 wallets sharing identical 4-char prefix and 3-char suffix pattern
  • ·Statistical impossibility for randomly generated addresses
  • ·All paying a single service (single-service concentration)
Buyers in cluster
17
Wash classification
100.0%

Signals matched

vanity_strictsingle_service
Operator Self-Testconfidence 0.66

Service C — Operator Self-Test

Pattern observed

  • ·Small cohort (n<10) of vanity-clustered wallets
  • ·Operator-controlled test traffic during service launch
  • ·Carved out as legitimate self_test, not classified as wash
Buyers in cluster
8
Wash classification
0.0%

Signals matched

vanity_broadsingle_servicesingle_txtiny_amount
Developer Dominanceconfidence 0.85

Service D — Developer Dominance

Pattern observed

  • ·102 distinct wallets, each paying a single service
  • ·Heavy bot pattern (top_svc_share ≥ 0.90)
  • ·Conservatively excluded from real_volume
  • ·May include legitimate production bots (operator self-disclosure pending)
Buyers in cluster
102
Wash classification
0.6%

Signals matched

single_service_concentrationhigh_tx_volumeregular_intervals
Clean Organic Serviceno wash flag

Service E — Clean Organic Service

Pattern observed

  • ·Diverse buyer base with varied transaction amounts
  • ·Distributed timing across the 30-day window
  • ·Buyers also use other unrelated services
  • ·100% organic_user classification
Buyers in cluster
45
Wash classification
0.0%

Signals matched

diverse_amountsdiverse_timingmulti_service_buyers

How wash detection works on x402watch

Open methodology, version-controlled in our repo. The pipeline runs the same five steps for every buyer, every day:

  1. 1.8-label classification in priority order (organic_user → verifier).
  2. 2.Cohort signals (uniform_amount, coordinated_start, uniform_tx_count).
  3. 3.Vanity clustering (strict + broad address-pattern matching).
  4. 4.Conservative developer label — only with strong recurrence evidence.
  5. 5.Real volume reporting excludes self_test, suspected_wash, and developer traffic.
Read full methodologyGitHub repo

For operators

Are you the operator of a service classified here?

Labels are deterministic and reproducible from public on-chain data. Open a GitHub issue with the seller address and we will walk through the signals that triggered the label.