What Are Entity-Adjusted Metrics?
Entity-adjusted metrics are on-chain statistics that have been corrected for the fact that a single blockchain address doesn't necessarily represent a single person or organization. Instead of counting raw wallet addresses, entity-adjusted analysis groups addresses controlled by the same actor — an exchange, a fund, a whale with 40 cold wallets — into one "entity" before running the numbers.
Why does this matter? Because raw address counts lie. Ethereum's blockchain doesn't know that Binance operates thousands of deposit addresses, or that a single trading desk splits its holdings across thirty wallets to obscure position size. Without adjustment, a metric like "active addresses" can look like it's showing broad-based network growth when it's really just one custodian shuffling funds between its own hot wallets.
I've seen this trip up plenty of retail traders who see a chart labeled "daily active addresses spiking 300%" and assume organic adoption, when in reality it's a single exchange migrating cold storage or an airdrop farm spinning up sybil wallets. Entity adjustment exists to strip out that noise.
How Entity Adjustment Works
The process generally follows three steps:
- Heuristic clustering — Analysts use behavioral patterns (common-input-ownership, deposit/withdrawal timing, gas-payer relationships) to flag addresses that likely belong to the same controller. This is the same logic behind wallet clustering techniques used to track whale coordination.
- Labeling known entities — Exchange hot wallets, known custodians, bridge contracts, and DeFi protocol treasuries get tagged using public labels (Etherscan, Arkham, Nansen all maintain these databases).
- Recomputation — Metrics like active addresses, holder count, or transaction volume are recalculated using entity counts instead of raw address counts.
Glassnode and CryptoQuant popularized this approach for Bitcoin and Ethereum data years ago, and it's now standard practice for any serious on-chain research. Their entity-adjusted "active addresses" figures are often 20-40% lower than raw address counts during periods of heavy exchange or bot activity — a meaningful gap if you're using the metric to gauge genuine demand.
Common Applications
| Raw Metric | Problem Without Adjustment | Entity-Adjusted Fix |
|---|---|---|
| Active addresses | Counts exchange internal transfers as "activity" | Merges exchange wallets into one entity |
| Holder count | Whale with 50 wallets looks like 50 holders | Clusters wallets by ownership heuristics |
| Transaction volume | Self-transfers between own wallets inflate volume | Filters out intra-entity transfers |
| NUPL / realized cap | Skewed by custodial reshuffling | Uses entity-level cost basis |
This matters most for narratives around decentralization and distribution. A token that appears to have 50,000 unique holders might really have 12,000 distinct entities once you cluster wallets tied to the same exchange custody system or the same farming operation — a distinction that changes how you read governance token concentration risk or Sybil-driven airdrop metrics entirely.
Myth vs Reality
Myth: Entity adjustment is a niche academic exercise that doesn't affect trading decisions.
Reality: It directly impacts how you interpret supply distribution, apparent demand growth, and even airdrop eligibility gaming. Projects have had to redesign token distribution criteria after entity-adjusted analysis revealed that a large share of "unique wallets" claiming an airdrop were controlled by a handful of farming operations — a pattern covered in token airdrop criteria gaming.
Myth: Entity clustering is 100% accurate.
Reality: It's heuristic-based and probabilistic. Sophisticated actors use techniques like CoinJoin, cross-chain bridging, or fresh wallet generation specifically to defeat clustering. Entity-adjusted numbers are a better estimate, not a perfect ground truth.
Treat entity-adjusted metrics as directionally correct, not gospel. Clustering heuristics improve constantly, but they'll never achieve 100% precision against a determined actor trying to fragment their footprint.
Why Traders Should Care
If you're using on-chain metrics for predicting token unlocks or trying to gauge real accumulation versus wash activity, raw address counts will mislead you. Entity-adjusted data gives a cleaner read on:
- Genuine new user growth vs. bot/Sybil noise
- True holder concentration (useful for assessing dump risk)
- Whether "active address" spikes reflect organic demand or custodial reshuffling
Platforms like Glassnode and Nansen publish entity-adjusted dashboards precisely because raw blockchain data, taken at face value, overstates decentralization and activity. For a broader primer on reading these signals correctly, see the guide on how to read and interpret on-chain metrics for trading.