Key points
- Holder percentages count balances or accounts; they do not automatically identify independent people.
- Classify pools, custodians, vesting contracts and unknown wallets before making concentration claims.
- On Solana, multiple token accounts can share one token authority; a largest-accounts list is not an owner census.
- Use a stated supply denominator, retain exclusions and distinguish transaction evidence from common-control inference.
A huge balance tops the token holder concentration screen. Before calling it a whale, open the address.
Token holder concentration measures how much of a token’s supply sits in a chosen set of addresses or accounts. It does not automatically measure how many independent people control that supply. A pool can rank first. An exchange can combine customer assets. One controller can divide a position across wallets.
The useful question is specific: which balances could be moved, by whom, under what conditions? This guide covers ERC-20 tokens on EVM chains and SPL tokens on Solana. It does not rate a particular project or set a universal safe percentage.
Token holder concentration starts with the denominator
Confirm the chain and full contract or mint address. A matching ticker is insufficient; use Panther’s contract-address verification guide before comparing lists.
Record the observation time, block or slot when available, supply figure and balances. Avoid combining yesterday’s supply with today’s holder list. Write the calculation explicitly:
Concentration = selected balances ÷ stated supply denominator × 100.
For an ERC-20 token, the standard defines totalSupply and balanceOf. Normalize raw values consistently using the token’s decimals. On Solana, the mint account stores supply and decimals. Neither an original launch allocation nor a vendor’s circulating-supply estimate should silently replace current on-chain supply.
If you exclude balances, publish the exclusions and keep the unadjusted result alongside them. Otherwise two percentages can describe different things while appearing directly comparable.
Classify the biggest entries before counting whales
Etherscan’s token-page documentation describes the Holders tab as a ranking by quantity or percentage. Read the address page behind each large entry, then keep a short evidence ledger.
| Entry | Evidence to seek | What remains unresolved |
|---|---|---|
| DEX pool | Correct pair or pool contract, token addresses and protocol records | Who controls liquidity positions and whether liquidity can be removed |
| Exchange custody | Explorer attribution corroborated by the exchange’s published address information | Customers’ beneficial balances and the exchange’s own inventory |
| Vesting or treasury contract | Verified implementation, beneficiary, release terms and privileged roles | Whether upgrades or exceptional paths can change those restrictions |
| Claimed burn address | Exact mechanism, successful transaction and supply effect | Whether a label alone establishes inaccessibility |
| Ordinary or unknown address | Balance, transfers and any independently supported attribution | Whether it shares control with other addresses |
A label is a lead for classification. Etherscan explains that name tags and labels provide attribution or category context, and offers a process for correcting false tags. Keep the supporting source with the label rather than treating the text as conclusive ownership evidence.
For a concrete pool example, Uniswap v2 documentation explains that providers receive LP tokens representing their contribution and redeem them for underlying liquidity. A token balance sitting in that pool is therefore different from an ordinary wallet’s immediately transferable stash. Classifying it as a pool does not establish that the liquidity is locked. Check liquidity withdrawal rights separately.
On Solana, token accounts are another layer
Solana’s token-account model allows a wallet to own multiple token accounts for the same mint. A token account has a mint, balance and token authority. Its token-account owner is different from the base account’s program owner, which identifies the Token Program.
This distinction matters when a tool returns accounts rather than aggregated owners. The official getTokenLargestAccounts method returns the 20 largest token accounts, not the 20 largest people or a complete ownership census.
Inspect each account’s token authority, aggregate accounts with the same authority where appropriate, and retain program-controlled custody as its own category. Aggregating only the returned 20 accounts can still miss an owner whose balance is spread across smaller accounts outside that list. Check whether an explorer already performs aggregation before adding numbers again.
A shared on-chain authority supports account-level grouping. It still does not tell you how many customers stand behind a custodial service.
Worked example: 12% and 15% can both be correct
Illustrative arithmetic only. These balances, classifications and burns are invented to explain the method; they are not observations of a live token, real wallets or a transaction test.
Assume an SPL token originally had 1 billion units. A genuine burn reduced its current mint supply by 200 million, leaving 800 million. Solana’s burn documentation confirms that burning reduces both the account balance and mint supply. Do not subtract the burned amount again from the current supply.
Suppose the current balances, already aggregated by token authority, are:
| Classified entry | Units | Share of current 800M supply |
|---|---|---|
| Verified pool custody | 300M | 37.5% |
| Verified exchange custody | 200M | 25% |
| Wallet A | 120M | 15% |
| Wallet B | 80M | 10% |
| All remaining addresses | 100M | 12.5% |
| Total | 800M | 100% |
Wallet A is 12% of the original billion and 15% of current supply. Wallets A and B together hold 200 million: 20% of original supply or 25% of current supply.
The table does not establish that A and B share control. Nor does the 62.5% in pool and exchange custody establish two whales. Those are separate attribution questions.
Removing the pool and exchange from your selected wallet group may clarify a particular question. Removing them from the supply denominator creates a different metric. Name it, show the calculation and explain why it answers your question; do not present it as the unchanged top-holder percentage.
A cluster is a hypothesis until the evidence supports control
Matching purchase times, similar trade sizes or a shared funding source can justify investigating a relationship. Each can also have another explanation, including common services. Preserve the transactions behind a proposed cluster and state the inference precisely.
A defensible note might say: “These addresses received funds from the same source; common control is unconfirmed.” Calling them the developer’s wallets requires stronger evidence. Panther’s wallet-tracing guide explains how to follow transfers and where attribution reaches a boundary.
Avoid counting a labeled cluster once and then adding its constituent wallets again. Equally, do not infer decentralization just because every individual wallet falls below a chosen threshold.
What to write down before acting on a holder warning
Save the chain, full token address, observation time, data source, denominator, selected balances, classifications and unresolved assumptions. For each exclusion, retain an evidence link. Recheck the largest entries when the snapshot is stale or material transfers have occurred.
For reflection or other non-standard ERC-20 tokens, explorer displays can diverge from contract state. Etherscan documents these discrepancies and describes querying balanceOf through Read Contract to check an address’s balance. A transfer-history sum is not always an adequate substitute.
Finally, keep distribution separate from execution risk. A large balance does not establish how much it could sell at the displayed price; that depends on available liquidity and trading restrictions. A dispersed list also says little about minting powers or contract privileges. Holder concentration is one evidence layer, not a safety certificate.
Sources and verification
Reviewed November 14, 2026 against the linked ERC-20 specification, Etherscan explanations, Solana documentation and Uniswap v2 pool documentation. The tables and calculations are Panther’s teaching examples. No real-wallet attribution, live token audit or executed transaction is claimed. Explorer layouts and labels may change; preserve the data behind any real assessment.
Sources
- 01ERC-20: Token Standard · Ethereum Improvement Proposalseips.ethereum.org/EIPS/eip-20
- 02Understanding the Token page · Etherscaninfo.etherscan.com/understanding-token-page/
- 03Public Name Tags, Labels & Public Notes · Etherscaninfo.etherscan.com/public-name-tags-labels/
- 04Pools — Uniswap v2 · Uniswapdevelopers.uniswap.org/docs/protocols/v2/concepts/pools
- 05Assets on Solana: mint and token accounts · Solanasolana.com/docs/tokens
- 06getTokenLargestAccounts · Solanasolana.com/docs/rpc/http/gettokenlargestaccounts
- 07Burn Tokens · Solanasolana.com/docs/tokens/basics/burn-tokens
- 08Common Issues with Non-Standard ERC-20 Tokens · Etherscaninfo.etherscan.com/non-standard-token-contracts/
Responses
No responses yet. Start the conversation.
Panther Journal — The intelligence brief
Stay ahead of the story.
Free. New investigations delivered when published. Unsubscribe anytime.






