Holder concentration: adjust for custody and contracts
Holder concentration begins as a distribution of balances across addresses, not a distribution of wealth across people. Large custodial addresses, application contracts, bridges, and treasuries can represent very different relationships. Preserve the raw address measure and show any entity adjustment as a separate, evidence-dependent interpretation.
The ERC-20 standard provides a balance interface for an address. It does not define beneficial ownership groups. A holder table built from balances therefore needs additional evidence before it can support claims about the number of independent economic owners.
Keep raw and adjusted views side by side
First define the asset contract, network, block, and denominator. Decide whether the denominator is total supply or a specifically defined circulation estimate. A top-ten share can change materially depending on that decision. Put the denominator directly in the metric label.
Build the raw ranked balance table before applying exclusions. Classify large entries by operational role only where evidence supports the classification. Keep unknowns as unknowns. Do not remove a large address merely because its size makes the chart look alarming or because someone informally calls it a contract.
| View | What it measures |
|---|---|
| Raw address share | Balances held by the largest identifiers |
| Custody-adjusted view | Estimate under stated custody assumptions |
| Treasury-separated view | Distribution outside specified treasury holdings |
| Uncertain-cluster range | Result under alternative grouping assumptions |
| Beneficial owners | Requires evidence not supplied by balances alone |
The adjusted views should be reproducible from the raw table, with every exclusion or grouping listed. Otherwise the adjusted number conceals the most important research decisions.
Hypothetical custody adjustment
Assume a fictional token has 100 million units under the selected denominator. One custodial address holds 30 million, a treasury holds 20 million, and the remaining addresses hold 50 million. The largest address share is 30%. That statement is correct as an address statistic.
If the custodian holds for many customers, 30% is not automatically one investor’s economic ownership. Conversely, treating the entire custodial balance as broadly distributed would also be unsupported without customer-level information. Report the custodial role and the missing breakdown rather than choosing whichever ownership story seems convenient.
Now suppose three addresses each hold five million and a researcher suspects one common owner. The combined 15% share is a scenario unless ownership evidence supports it. Show results under separate-address and grouped-address assumptions. The hypothetical example illustrates uncertainty bounds, not an estimate for any actual token.
Inspect restrictions separately
Vesting contracts, bridge custody, and treasury accounts can affect the interpretation of concentration. Their balances may be large while the underlying rights are distributed or restricted in different ways. Record the control and entitlement question for each material category rather than assigning a single universal adjustment.
Do not mix transferability with ownership concentration. An address can hold units subject to a restriction, and an unrestricted balance can be controlled by an entity with many private obligations. If the research question concerns possible supply availability, add a separate restrictions analysis rather than assuming concentration alone answers it.
For changes over time, keep the same classification rules or provide a restatement bridge. A newly labeled custody address can lower an adjusted concentration estimate without any tokens changing hands. That is a knowledge revision and should not be described as decentralization caused by market distribution.
Use the result within its limits
Concentration can help identify dependencies worth investigating: custody exposure, treasury control, governance participation, or sensitivity to a small set of addresses. It does not establish coordinated behavior, willingness to sell, or future market impact. Those are separate claims requiring additional evidence.
The method is unsuitable for a precise person-level ownership count when custody and address clustering are unresolved. Common mistakes include equating one address with one person, excluding all contracts indiscriminately, grouping addresses through weak transfer patterns, and changing the denominator between periods. A useful conclusion states the raw result, the adjustments, and which uncertainty materially changes interpretation.
FAQ
Should exchange addresses be removed from holder rankings?
Keep the raw ranking. If an adjusted view is useful, disclose the custody assumption and avoid implying that the unknown customer distribution has been measured.
Does a concentrated holder table prove coordinated selling risk?
No. It identifies balance concentration under a rule. Coordination, intentions, restrictions, and market access require separate research.