When an entity spreads its funds across many wallets, the number of on-chain holding addresses simply multiplies by the number of splits – break into 100 addresses, and the holder count falsely grows 100 times. This is extremely common when project teams pump their data to create fake popularity. What looks like "100k holders" might be an illusion created by just a few hundred real users splitting tokens repeatedly. Below we explain the inflation mechanism, how to spot it, and how to reach a more realistic number once you have the holder list.
Why Splitting Wallets Inflates Holder Counts
On-chain statistics define a "holder" (or "Unique Address") by address alone, not by real people. If a project team or a whale spreads 1 million tokens evenly into 500 new addresses, block explorers will report 500 additional holders. This number has nothing to do with genuine unique users.
Inflation multiplier = total number of split addresses ÷ number of real entities. In extreme cases, a single entity can create tens of thousands of "holders." Block explorers like Etherscan and BscScan do not automatically deduplicate – the dashboard simply inflates.
⚠️ Risk Warning: Controlling many addresses to create fake holdings may constitute market manipulation or false statements in many jurisdictions. For example, the U.S. SEC has sued ICO projects that used multiple wallets to fake volume, leading to fines and market bans for main operators. For ordinary investors, equating a high holder count with "strong community" and buying in can easily mean buying right before a pump-and-dump exit.
Step 1: Get a Complete List of Token Holder Addresses
First, you need the top N holder data for the target token contract. No coding required – you can do it with a block explorer or a free API.
Goal: Obtain all addresses that currently hold the token, along with their balances.
How to do it:
- Option A: Check directly on Etherscan / BscScan
Open the token contract page → Click the "Holders" tab. By default you see the top 10,000 addresses, with pagination. For more, switch to the API.
- Option B: Pull data via the block explorer's free API
For BscScan, use the token/tokenholderlist endpoint. You'll need a free API key. Example request:
https://api.bscscan.com/api?module=token&action=tokenholderlist&contractaddress=&page=1&offset=10000
Repeat pagination until you collect all addresses. Etherscan, PolygonScan and others work the same way.
Completion criteria: Export at least one CSV or table with the columns address and balance, covering 99%+ of the total circulating supply (the first ten pages usually capture the vast majority). If a token has over 100k holder addresses, you may not need them all; prioritize the top 10,000–20,000 addresses by balance, filtering out zero balances.
Common pitfalls: API responses may use scientific notation or trailing zeros; opening in Excel might truncate the balance. Set the balance column to text format before importing, or convert the raw on-chain decimals to a readable amount (e.g., BSC tokens often use 18 decimals; divide by 10^18).
Step 2: Identify Address Clusters Likely Controlled by One Entity
In a pumped holder list, many addresses share clear signs: funds coming from the same source, operations around the same time, and highly identical balances.
Goal: Use a few dimensions to flag suspicious address groups and pull out the hundreds or thousands of addresses controlled by one entity.
How to do it:
- Trace the funding source
Spot-check a batch of small holding addresses (e.g., balances between 100 and 1000 tokens). Look at the transaction in which they first received the token. If dozens of addresses all received their tokens from the same sender wallet within a short time window (a few hours), you can be nearly certain it's one entity splitting its position.
- Check the first interaction time
If a whole group of addresses first received the token on the same day within a two-hour window, combined with funding source, you can classify them as one cluster.
- Analyze the balance distribution
Sort balances in descending order. Many inflation schemes give "soldier addresses" exactly equal or nearly identical amounts – for example, all exactly 1000.00. Such mechanical equality is the mark of batch generation; natural retail holdings rarely look this tidy.
- Cross-reference other assets and behaviors
If those same addresses also hold a junk token used for "farming interactions," or they all called the same unverified contract function, the confidence of linkage is very high. Use the block explorer's "Token Transfers" tab to check which contracts the addresses have commonly touched in the past.
Completion criteria: Mark several address clusters, each with clear evidence of a common source transaction or behavior. Record the address count and the first distribution transaction hash for each cluster.
For batch processing, you can use platforms like Dune Analytics to write queries (basic SQL needed). For more on building on-chain monitoring dashboards, see the article On-Chain Data Query in Practice.
Step 3: Estimate the True Number of Unique Holders and the Inflation Multiplier
Once you have the cluster data, deduplicate by counting each cluster as one "entity." Inflation multiplier = reported holder addresses ÷ deduplicated entity count.
Goal: Calculate the real number of controlling entities and compare it to the holder count shown by the explorer.
How to do it:
- Treat every address cluster confirmed in Step 2 as one entity.
- For remaining addresses that cannot be proven to be linked, tentatively count each as one entity.
- Sum all entities to get the minimum number of unique entities. Note that this still might overestimate the true number, because some undiscovered clusters and individual users with multiple wallets remain.
Example calculation:
Suppose a token shows 20,000 holding addresses.
You discover:
- Cluster A (same source, same time, equal 1,000 balances): 3,500 addresses → count as 1 entity.
- Cluster B (same source distribution, 10,000 balances each): 1,200 addresses → count as 1 entity.
- Remaining 15,300 addresses show no clear link → tentatively 1 entity each.
Minimum entities = 1 + 1 + 15,300 = 15,302.
Inflation multiplier ≈ 20,000 ÷ 15,302 ≈ 1.3x. If Cluster A belongs to the project team and there are more undetected clusters, the real inflation can be even higher.
Completion criteria: Output a deduplicated entity count and the minimum inflation multiplier. Even if you can't find every cluster, simply removing the most obvious same-source distribution groups already reveals how much the data is watered down.
Common Questions
Does a rising holder count always mean fake inflation?
Not always. If a token expands naturally after an airdrop or a major exchange listing, addresses rarely share the same source and the growth curve is smoother. A 3,000% spike in one day with many equal balances almost certainly points to artificial splitting.
Are all equal balances the result of manipulation?
No. Some staking contracts or dividend mechanisms cause many addresses to hold equal shares. Those addresses usually interact frequently with the protocol contract. You can distinguish them by checking whether they called standard staking methods. Artificially inflated addresses typically only receive tokens and do nothing else.
Block explorers include exchange and contract addresses in the "Holder" count – how to handle them?
Many explorers already exclude known exchange wallets and contract addresses from holder statistics, e.g., Etherscan filters out the 0x0 address and some contracts. However, the rules are not fully transparent. After exporting addresses, manually remove known centralized exchange deposit addresses, burn addresses, and black hole addresses before clustering. This avoids misclassifying them as fake holders.
Can data be restored after a project withdraws the inflated addresses?
On-chain records are permanent. The history of an artificially inflated holder count is captured by some on-chain analytics platforms (e.g., historical snapshots on Nansen or CoinMarketCap). Mainstream explorers, however, only show the current state. If you need historical data, you can query balance tables at a specific block height on Dune and reconstruct the true holder picture at that past moment.
Verification after the analysis: Open the token page you just examined, check the address list under the "Holders" tab, and randomly pick 10 addresses. One by one, look at their first receiving transaction. If over 7 out of 10 came from the same distribution address you flagged, your clustering is solid. With practice, an initial screening of one token takes about 30 minutes – enough for you to judge how much substance is behind that "number of holders."
