Crowd-traced exchange addresses

Mapping the exchanges, one tracer at a time

A unique native token deposited to an exchange behaves like a tagged banknote: it cannot commingle the way ada does, so it can be followed hop by hop from a deposit address into the exchange’s internal wallets. Each depositor reports which exchange they used; that self-report is the only off-chain input. For each cluster we publish the full tracer-label distribution, and the agreement among independent depositors, plus whether the address looks like exchange infrastructure, sets its confidence. How we keep the data honest →

Mockup
Low-fidelity mockup

A draft sketch of what the study will deliver, filled with invented data. No real exchange has been labelled yet; the layout, columns and label-distribution below are what the live map will show, populated from on-chain tracer walks once the study runs.

Example cluster map · wireframe
ABCExchange Aconfidence: highExchange Bconfidence: mediumExchange Cconfidence: lowcold · corehot · middle ringdeposit · edgegray = unmapped
Exchange Alphaconfidence: high12 depositors
AddressRoleTracer labelsDepositorsShapeConfidence
addr1q9x7e3k…c9d0ecold / at rest9× Alpha1× Beta10842,133 · 5,120 fdrhigh
addr1q8h2alp…9d0e1hot / consolidation6× Alpha6210,044 · 1,830 fdrhigh
addr1q7g3alp…0b1c2hot / consolidation4× Alpha498,011 · 940 fdrhigh
addr1q6f4alp…1a2b3deposit address2× Alpha23 · 2 fdrmedium
Exchange Betaconfidence: medium6 depositors
AddressRoleTracer labelsDepositorsShapeConfidence
addr1q5e9bet…3b2a1cold / at rest5× Beta5412,900 · 3,110 fdrmedium
addr1q4d8bet…3f2e1hot / consolidation3× Beta377,002 · 610 fdrmedium
addr1q3c7bet…2c1b0deposit address2× Beta22 · 2 fdrlow
Exchange Gammaconfidence: low2 depositors
AddressRoleTracer labelsDepositorsShapeConfidence
addr1q2b6gam…1b0a9hot / consolidation2× Gamma21,844 · 70 fdrlow
addr1q1a5gam…0a9z8deposit address1× Gamma11 · 1 fdrlow
Mapper leaderboardranked by agreed nodes mapped8 contributors
#Study walletAgreed nodesBranches
1addr1qxmapp…e1f24114
2addr1qymapp…e2f33812
3addr1qzmapp…f3g43311
4addr1qwmapp…g4h5279
5addr1qvmapp…h5j6228
6addr1qumapp…j6k7196
7addr1qtmapp…k7l8145
8addr1qsmapp…l8m993

Only corroborated nodes score, so the board rewards mapping the wider set agrees on, not raw deposit volume or lone claims. Spreading a batch across deposit addresses and over time is how you climb it.

Rejected / withheld · the poisoning defence

False or thin claims do not silently vanish; they stay visible and explained:

  • addr1q0z4sme…4z3y2 claimed a major exchange
    one report, on an address with no exchange shape (no sweep, a single funder) and no corroboration → withheld as a likely false attribution.
  • addr1q9x7e3k…c9d0e claimed Exchange Beta
    one tracer claimed Beta on a cluster nine others labelled Alpha → outlier in the distribution, not published as Beta.

47example tracers walked. “Shape” is the address’s hub-degree (lifetime receives · distinct funders): a high count is the exchange-vs-closed-wallet signal. Every figure on this page is invented for illustration.

An exchange label is an inference from voluntary deposits, on-chain footprint only, not a confirmed entity. Reaching an address shows contact and sweep behaviour, not custody of any named party’s funds. What this proves, and what it doesn’t →