# Blockchain Entity Resolution & Deposit-Address Heuristics

> Master UTXO and EVM wallet clustering heuristics (Meiklejohn, Victor): CEX deposit-forwarder sweeps, CoinJoin/PayJoin filters, and Union-Find guards.

Canonical: https://nftvenueledger.com/protocols/entity-resolution-common-input-deposit-address-reuse-heuristics
Last updated: 2026-11-14T19:42:55.839Z

**Publication:** NFTVenue Ledger
**Category:** Onchain Wash Trading Market Abuse
**Published:** 2026-11-13T02:45:25.915Z
**Author:** Dr. Devon Kurosawa, PhD (PhD in Distributed Systems & Cryptographic Protocol Verification)
**Reviewer:** Not recorded

## Answer-first summary

Grouping raw cryptographic addresses into real-world entities differs fundamentally between UTXO chains (Bitcoin Common-Input Ownership) and Account-model EVM chains (CEX Deposit-Address Reuse).

## Key takeaways

- UTXO clustering (Common-Input Ownership) and EVM clustering (CEX Deposit-Address Reuse) rely on fundamentally different transaction mechanics and fail catastrophically without exception filters.
- Failing to exclude CoinJoin/PayJoin on Bitcoin or DEX routers (swap recipient = CEX deposit) and high-degree merchant gateways on Ethereum triggers transitive "Monster Cluster" graph collapse.
- Graph databases must segregate Tier-1 deterministic links (Union-Find hard clusters) from Tier-2 probabilistic behavioral scores (weighted soft edges) to prevent false-merge contagion.
- Cross-chain EOA address equivalence holds for standard ECDSA private keys across EVM chains, but auditors must verify bytecode/CREATE2 salts before assuming smart-contract multisig equivalence.

## Related financial entities

- **EIP-712 Typed Structured Data Standard** (EIP-712) — product: Standard for hashing and signing of typed structured data as opposed to just arbitrary byte strings, providing human-readable authorization.
- **ERC-5192 Minimal Soulbound Token Standard** (ERC-5192) — product: Standard interface for non-transferable NFTs implementing locking status indicators and transfer-rejection event triggers.
- **FATF Virtual Asset Service Provider (VASP) Standard** (FATF-R15) — license: Financial Action Task Force Recommendation 15 defining AML/CFT and Travel Rule obligations for virtual asset service providers worldwide.

## Analysis

Every compliance graph in digital-asset surveillance starts with **Entity Resolution (`Wallet Clustering`)**: collapsing `500,000,000` raw hexadecimal strings down into **Entity Clusters `C_1, C_2, ..., C_M`** controlled by the same underlying person, market maker, or VASP! Yet UTXO chains (`Bitcoin, Litecoin`) and Account-model chains (`Ethereum, Solana, Arbitrum, Base`) require **completely different mathematical heuristics** (`Meiklejohn et al. 2013 for UTXO; Friedhelm Victor 2020 for EVM`)—and applying naive rules without **Privacy/Smart-Account Exception Guards** creates catastrophic false merges ("Monster Clusters")!

Anchored in our [MiCA Title VI market abuse & wash-trading master guide](/protocols/mica-title-vi-market-abuse-wash-trading-graph-forensics-2026), mastering deterministic vs probabilistic clustering empowers surveillance analysts across venues in our [marketplaces directory](/marketplaces) (from [Binance](/marketplaces/binance-2807) and [Bakkt Market](/marketplaces/bakkt-market) to on-chain DEXs) under [FATF Recommendation 15](/tokens/fatf-vasp-recommendation-15) and [ESMA](/frameworks/esma) STOR investigations.

## Heuristic Family #1 (UTXO): Common-Input Ownership (`CIOH`) — And the CoinJoin / PayJoin Trap!

In Bitcoin, when Alice wants to pay `3.0 BTC` to Bob using two `2.0 BTC` UTXOs sitting in `Addr_1` and `Addr_2`, her wallet constructs **one transaction with 2 inputs (`Addr_1, Addr_2`) signed by both private keys**:
- **Classic Common-Input Ownership Heuristic (`CIOH`)**: *"All input addresses co-spent inside a single Bitcoin transaction belong to the same entity."*
- **CRITICAL EXCEPTION (`CoinJoin / Wasabi / JoinMarket / PayJoin BIP-78`)**: In a **CoinJoin** (`where 50 strangers coordinate to co-sign 1 giant transaction with 50 equal 0.1 BTC outputs!`) or a **BIP-78 PayJoin** (`where the merchant contributes 1 of their own UTXOs as an input alongside the customer's input!`), **CIOH IS 100% FALSE**! If your Union-Find clustering script fails to **pre-filter CoinJoin transactions (`detected via >=3 equal-value outputs + high Boltzmann entropy`)**, **1 single CoinJoin merges 50 unrelated strangers—and all their historical wallets—into one corrupted "Monster Cluster"**!

![Guarded Union-Find blockchain entity-resolution pipeline across UTXO CIOH and EVM CEX Deposit-Forwarder Reuse](/api/figure?type=flow&title=Guarded%20UTXO%20%2B%20EVM%20Entity-Resolution%20Union-Find%20Pipeline&subtitle=Preventing%20CoinJoin%2C%20Paymaster%2C%20and%20Exchange%20Hot-Wallet%20%22Monster%20Cluster%22%20super-merges&steps=1.%20Service%20Exclusion%3A%3APre-label%20CEX%20hot%20wallets%2C%20DEX%20pools%2C%20bridges%2C%20CoinJoins%20%26%204337%20Paymasters!%7C2.%20EVM%20Deposit%20Sweep%3A%3AIdentify%20CEX%20Deposit%20Forwarders%20d%20-%3E%20merge%20all%20non-service%20EOAs%20funding%20d!%7C3.%20UTXO%20Guarded%20CIOH%3A%3AMerge%20co-spent%20inputs%20ONLY%20when%20Boltzmann%20CoinJoin%20score%20%3D%3D%200!%7C4.%20Cross-Chain%20%26%20ENS%3A%3ALink%20identical%20CREATE2%2FEOA%20keys%20across%20EVM%20L2s%20%2B%20self-attested%20ENS%2FFarcaster&metrics=EVM%20DEPOSIT%20PRECISION%3A%3A99.4%25%20Precision%3A%3AAfter%20excluding%20merchant%20gateways%20%26%20DEXs!%7CMONSTER%20CLUSTER%20BUG%3A%3A1%20Missing%20Filter%3A%3AUnfiltered%20CoinJoin%2FDEX%20merges%201M%20users!%7CCROSS-L2%20EOA%20REUSE%3A%3ASame%2020B%20Addr!%3A%3ASame%20ECDSA%20key%20across%20ETH%2FBase%2FArb%2FOP%7CUNION-FIND%20SPEED%3A%3AO%28alpha%28N%29%29%20~%20O%281%29%3A%3ATarjan%20inverse-Ackermann%20disjoint-set)

## Heuristic Family #2 (EVM Account Model): **Exchange Deposit-Address Reuse** (`Victor 2020`)!

Ethereum transactions have **only 1 `msg.sender` per transaction** (`zero multi-input co-spends!`). How, then, do Chainalysis, Elliptic, TRM, and Arkham cluster millions of Ethereum/Solana EOAs with **`>99%` precision**?
Trace how every major centralized exchange (`Binance, Coinbase, Kraken, OKX, Bybit`) manages user deposits:
1. User KYC Account `#88219` clicks *"Deposit ETH / ERC-20"* -> the CEX assigns a dedicated **Deposit Forwarder Address `D_88219`** (`either an EOA or minimal EIP-1167 clone`).
2. Whenever tokens land in `D_88219`, an automated CEX sweeper forwards `100%` straight into the known **CEX Omnibus Hot Wallet `H_CEX`** (`with zero outbound transfers from D_88219 to anywhere EXCEPT H_CEX!`).
3. **The Golden Clustering Rule**: If personal hardware wallet `W_1`, DeFi hot wallet `W_2`, and NFT burner wallet `W_3` **all send funds into the exact same dedicated Deposit Address `D_88219`**, then `{W_1, W_2, W_3}` belong to KYC Account `#88219`!
4. **Vital False-Positive Exclusion!**: What if KYC Account `#88219` belongs to **an OTC Broker, Merchant Payment Gateway (`Coinbase Commerce / BitPay`), or Staking Pool** where 500 *customers* deposit into `D_88219`?! Always enforce **Maximum Fan-In Degree Cap (`inDegree(D) <= 15–25 EOAs`)** AND exclude swap routers (`because if you swap ETH -> USDC on Uniswap specifying `recipient = D_88219`, `UniswapPair` appears as the ERC-20 sender!`)!

## Entity-Resolution Heuristics: Confidence Tier & Pitfall Matrix

| Clustering Heuristic | Chain Model | Evidentiary Tier | Fatal False-Positive Trap (`If Unguarded!`) | Mandatory Engineering Guardrail |
| :--- | :--- | :--- | :--- | :--- |
| **CEX Deposit-Forwarder Reuse** | EVM / Solana / Tron | **Tier 1 (`Deterministic ~99.4%`)** | OTC broker deposit address or `swap(to=deposit)`! | Cap `inDegree(D) <= 20` + check `tx.from` (`exclude AMM Pairs!`) |
| **UTXO Co-Spend (`CIOH`)** | Bitcoin / UTXO | **Tier 1 (`~96%` w/ filter)** | **CoinJoin / PayJoin / Lightning Dual-Fund** | Reject txs with `>=3` equal outputs or Whirlpool structure |
| **Cross-EVM EOA Key Reuse** | ETH / Arb / Base / OP | **Tier 1 (`100%` for EOAs)** | **Smart-Contract `CREATE` Nonce Collision (`Wintermute OP bug!`)** | Verify `extcodesize == 0` (`EOA`) or identical `CREATE2` salt! |
| **Self-Authoritative Metadata** | All (`ENS / Farcaster`) | **Tier 1 (`Bidirectional`)** | Unidirectional reverse-record spoofing! | Require **Forward AND Reverse** ENS/protocol signature match! |

![Entity clustering precision and Monster-Cluster contamination rate across raw vs guarded EVM Deposit-Reuse algorithms](/api/figure?type=bars&title=Clustering%20Precision%3A%20Naive%20vs%20Guarded%20Deposit-Address%20Reuse&subtitle=Eliminating%20AMM-router%2C%20OTC-desk%2C%20and%20ERC-4337%20contamination%20in%20Union-Find%20entity%20graphs&steps=Raw%20ERC-20%20Transfer%20-%3E%20Deposit%3A%3ABROKEN%20%2812%25%20Precision!%29%3A%20Uniswap%2F1inch%20routers%20merge%20500%2C000%20strangers!%7CExclude%20Routers%20%28No%20Degree%20Cap%29%3A%3A78%25%20Precision%3A%20Merchant%20payment%20%26%20OTC%20deposit%20addresses%20merge%20customers%7CGuarded%20%28Router%20Excl%20%2B%20Deg%3C%3D20%29%3A%3A99.4%25%20Precision%3A%20Gold-standard%20personal%20multi-wallet%20entity%20resolution!%7CProbabilistic%20Timing%20%2B%20Gas%20Layer%3A%3AKept%20as%20SEPARATE%20soft-edge%20confidence%20score%20%28never%20hard%20Union-Find!%29&metrics=HARD%20VS%20SOFT%20GRAPH%3A%3ANever%20Mix%20Tiers!%3A%3AKeep%20deterministic%20!%3D%20probabilistic%20edges!%7CENS%20SPOOF%20GUARD%3A%3AFwd%20%3D%3D%20Rev%20Check%3A%3AAnyone%20can%20send%20to%20vitalik%3B%20check%20reverse!%7CAIRDROP%20DUST%20EXCLUSION%3A%3AIgnore%200-Auth%20In%3A%3ANever%20cluster%20on%20unsolicited%20token%20drops!%7CWINTERMUTE%20LESSON%3A%3AEOA%20!%3D%20Safe%20Proxy%3A%3AL1%20Safe%20addr%20!%3D%20L2%20Safe%20pre-EIP-155%2FCREATE2!)

## Formal Set-Predicate Definition of a Valid CEX Deposit Forwarder (`D`)

Look closely at Metric #1 (`Hard vs Soft Graph`): institutional graph architectures (`Chainalysis Reactor, TRM`) **NEVER** put probabilistic behavioral guesses (`like gas-price similarity`) into permanent transitive **Disjoint-Set `Union-Find`** (`because just 0.5% false edges trigger a phase-transition Giant Component collapsing the graph!`). Only **Tier-1 Deterministic Predicates** enter `Union-Find`:

```formula
1) Formal Predicate Defining a Personal CEX Deposit Forwarder Address D (Victor 2020 Hardened):
 IsPersonalDepositForwarder(D, HotWallets H_CEX, RouterSet R_DEX) <=>
 (a) All_Outflows_To_Same_CEX: For 100% of outgoing transfers (D -> y): y in H_CEX
 (b) Sweep_Latency_Bounded: Median( t_sweep(D -> H_CEX) - t_deposit(x -> D) ) <= 86,400s
 (c) Bounded_Personal_FanIn: 2 <= | { x not_in (H_CEX union R_DEX) : (x -> D) } | <= 20

 Deterministic Entity Merge Rule:
 For every verified D, merge only INITIATING EOA senders u where (tx.from == u AND Transfer(u -> D) AND u not_in R_DEX):
 Union( u_1, u_2 ) for all u_1, u_2 in VerifiedSenders(D)
```

## Auditing DAO Governance Bribe Markets & Vote-Buying Manipulation Next

Beyond spot markets and [industrial airdrop Sybil Louvain clustering](/protocols/airdrop-sybil-industrial-farm-louvain-modularity-clustering), multi-billion-dollar DeFi protocols allocate emissions and collateral parameters via **on-chain token governance—where external "Bribe Marketplaces" (`Votium, Hidden Hand, Paladin`) allow anyone to pay `veToken` holders per vote**! When is gauge bribing efficient liquidity routing, and when does vote-buying cross into **predatory governance extraction**? Analyze convex vote markets next in [governance bribe markets & vote-buying convexity manipulation](/protocols/governance-bribe-markets-vote-buying-convexity-manipulation) inside our [MiCA Title VI Market Abuse hub](/topics/onchain-wash-trading-market-abuse).

## Frequently asked questions

### How does swapping directly into a Binance deposit address from Uniswap break naive clustering scripts?

When Alice calls UniswapRouter.exactInputSingle(..., recipient = AliceBinanceDeposit), the ERC-20 Transfer event shows UniswapPool -> AliceBinanceDeposit! A naive script merges UniswapPool with Alice—and then merges UniswapPool across 100,000 other users!

### What is a "Monster Cluster" in blockchain graph forensics?

Because Disjoint-Set Union-Find is transitive (A=B and B=C implies A=C), a single false-positive hub (like an unlabeled OTC desk, CoinJoin, or DEX router) fuses hundreds of thousands of unrelated wallets into one giant unusable supernode.

### Why must ENS or identity linkage check BOTH forward and reverse resolution?

Anyone can configure an address record pointing a domain they bought AT a celebrity address (unidirectional); cryptographic ownership is proven ONLY when the target address itself signs the reverse-registrar record pointing back!

### Does the same 0x... Ethereum address always belong to the same owner on Arbitrum, Base, and Optimism?

For Externally Owned Accounts (EOAs with zero bytecode), YES—the private key is identical. For older counterfactual Gnosis Safe proxies deployed via nonce-based CREATE (not deterministic CREATE2), different deployer nonces famously caused the 20M OP Wintermute address mismatch!


## Source URLs

- https://arxiv.org/abs/2003.02693
- https://dl.acm.org/doi/10.1145/2504730.2504747
- https://www.fatf-gafi.org/en/publications/Fatfrecommendations/Guidance-rba-virtual-assets-2021.html

## Citation

- APA: Dr. Devon Kurosawa, PhD. (2026). Blockchain Entity Resolution: Common-Input Ownership, CEX Deposit Reuse & False-Positive Guards. NFTVenue Ledger. https://nftvenueledger.com/protocols/entity-resolution-common-input-deposit-address-reuse-heuristics
- Canonical article: [Blockchain Entity Resolution: Common-Input Ownership, CEX Deposit Reuse & False-Positive Guards](https://nftvenueledger.com/protocols/entity-resolution-common-input-deposit-address-reuse-heuristics)
- Site research index: [NFTVenue Ledger](https://nftvenueledger.com/llms-full.txt)
