Credit Scoring

The core mechanism involves analyzing a wallet's complete, immutable transaction history. This includes:

• Volume and frequency of transactions across protocols.
• Counterparty diversity and interactions with known entities.
• Historical behavior patterns like liquidation events or successful repayments.
• Asset composition and holding duration (HODL behavior). This data forms the raw input for scoring algorithms, moving beyond self-reported data to observable on-chain actions.

A key Web3 advantage is that a credit score is a composable primitive. Once generated, it can be permissionlessly used by any integrated DeFi protocol without re-submission. This creates portability:

• A score from one lending platform can be used to access undercollateralized loans on another.
• It enables cross-protocol syndicated loans where risk is shared based on a verifiable score.
• Developers can build new financial products that ingest standardized score data feeds.

Scores are not static but are tied to programmable risk parameters set by protocols or communities. This allows for:

• Dynamic credit limits and loan-to-value (LTV) ratios that adjust based on score tiers.
• Automated interest rate curves where borrowers with higher scores receive better rates.
• Custom risk models for specific asset classes (e.g., NFT-backed lending vs. stablecoin pools).
• Governance-controlled parameter updates to respond to market conditions.

On-chain scoring must solve the Sybil attack problem, where a user creates many wallets to game the system. Solutions include:

• Graph analysis to cluster wallets controlled by a single entity.
• Proof-of-humanity or decentralized identity (DID) attestations as a foundational layer.
• Address linking via consistent funding sources or behavioral fingerprints.
• Reputation decay for inactive addresses to prevent score accumulation without ongoing activity.

Unlike opaque traditional credit scores, on-chain methodologies can be fully transparent and auditable. Features include:

• Open-source scoring algorithms where the weighting of factors is public.
• Verifiable data sources from public blockchains like Ethereum or Solana.
• Score provenance tracking, allowing users to see which transactions influenced their score.
• Community governance over model updates, moving control from private corporations to token holders.

Scores update in near real-time based on live on-chain activity, creating a dynamic financial reputation. This enables:

• Immediate impact of positive actions (timely repayments boost score).
• Rapid detection of risk (e.g., sudden high-leverage positions trigger score downgrades).
• Continuous underwriting where loan terms can be adjusted during the loan period based on behavioral changes.
• Micro-achievements where small, consistent positive actions contribute to reputation building.

The foundational layer of on-chain analysis, examining the complete ledger of wallet-to-wallet transfers and smart contract interactions. Key metrics include:

• Volume and frequency of transactions.
• Counterparty diversity (number of unique addresses interacted with).
• Transaction age and recency patterns.
• Gas fee payment behavior, indicating willingness to pay for priority.

Analysis of a wallet's on-chain portfolio at a specific block height or over time. This assesses financial stability and risk appetite.

• Total Value Locked (TVL) across DeFi protocols.
• Asset diversification across cryptocurrencies, stablecoins, and NFTs.
• Collateralization ratios in lending protocols like Aave or Compound.
• Long-term holding vs. speculative trading patterns.

Granular data from interactions with decentralized finance applications, providing deep insight into sophisticated financial behavior.

• Borrowing history: Loan amounts, repayment timeliness, and liquidation events on platforms like MakerDAO.
• Lending activity: Assets supplied and yield earned.
• Trading history: DEX swap volumes, slippage tolerance, and impermanent loss from liquidity provision.
• Governance participation: Voting weight and proposal involvement.

Data derived from a wallet's non-financial on-chain footprint, building a profile of trust and community standing.

• Soulbound Tokens (SBTs): Non-transferable credentials for attestations.
• DAO membership and contribution history.
• Gitcoin Grants donations and matching history.
• ENS domain ownership and age, signaling established identity.

Data sourced from native on-chain credit systems that explicitly track and score financial obligations.

• Repayment history from credit delegation pools (e.g., Aave's Credit Delegation Vaults).
• Credit limit utilization and default events from protocols like Cred Protocol or Spectral Finance.
• Under-collateralized loan performance from platforms like Goldfinch.

Analysis of when and how a wallet interacts with the blockchain, revealing behavioral consistency and strategic intent.

• Time-based analysis: Activity during bear vs. bull markets, or reaction to volatility.
• Sequence of actions: Complex multi-step DeFi strategies (e.g., leveraged yield farming).
• Address clustering: Linking multiple addresses (EOAs, smart contract wallets) to a single entity for a holistic view.

A conceptual implementation where a credit score directly enables uncollateralized or undercollateralized loans. This represents the end-goal for many on-chain credit systems. Key mechanisms include:

• Credit Limits: A smart contract sets a borrowing limit based on a verified credit score or reputation NFT.
• Default Consequences: Penalties can include score degradation, social enforcement, or legal recourse via on-chain legal frameworks like Kleros or Aragon.
• Protocol Examples: Early experiments include TrueFi (institutional underwriting) and Goldfinch (pool-based assessment), though they primarily use off-chain due diligence.

A Sybil attack involves creating many fake identities to manipulate a reputation or scoring system. In decentralized finance, this is a primary method to artificially inflate a credit score. Mitigation strategies include:

• Proof-of-Humanity or soulbound token (SBT) attestations to link identities.
• Analyzing transaction graph clustering to detect coordinated wallets.
• Implementing time-based scoring that requires sustained, legitimate activity over time.

The underlying scoring algorithm is a critical risk factor. A black-box model can fail unpredictably or be gamed. Key considerations are:

• Explainable AI (XAI): Ensuring score components are auditable and understandable.
• Parameter risk: Over-reliance on specific on-chain metrics (e.g., NFT floor prices) that can be volatile or manipulated.
• Governance: Who can update the model parameters, and what is the process for doing so?

While blockchain data is public, sophisticated credit models can infer sensitive financial behavior. This creates risks of:

• Wallet deanonymization through pattern analysis of transactions and holdings.
• Front-running opportunities if a score change is predictable, others may act on it before the user.
• Compliance with regulations like GDPR, which conflict with immutable public ledgers. Solutions include zero-knowledge proofs (ZKPs) for private computation on public data.

Credit-based lending often uses a hybrid model of reputation and collateral. This introduces complex risk interactions:

• Collateral volatility: A sharp drop in asset value can trigger liquidation before a score can adjust.
• Liquidation cascade: Mass liquidations in one protocol can depress collateral prices and trigger failures in connected credit systems.
• Recovery rate uncertainty: The actual value recovered from a defaulted, undercollateralized loan is highly uncertain on-chain.