Observed risk methodology

Observed risk definition

Observed risk summarizes available evidence. It does not predict probability of loss and is not an audit, guarantee, credit rating, or investment recommendation.

Risk dimensions

Protocol / smart contract

Contract age, audit evidence, bug bounty presence, exploit history, and identity verification.

Examples: Audit reports, deployment history, incident records.

Why it matters: Smart-contract flaws or unresolved incidents can cause loss even when rates look attractive.

Governance / admin control

Upgradeability, admin key type, timelocks, and pause authority.

Examples: Multisig admin, timelock duration, emergency pause scope.

Why it matters: Governance can change economics or halt markets after you deposit.

Oracle

Oracle dependence, provider identity, single-source risk, and incident history.

Examples: Chainlink feeds, custom oracles, recent oracle manipulation events.

Why it matters: Oracle failures or manipulation can break markets independent of core contract quality.

Asset

Stablecoin classification, depeg history, bridged representations, and volatile exposure.

Examples: Fiat-backed vs algorithmic stablecoins, bridged wrappers, multi-asset pools.

Why it matters: Asset risk can turn stable yield into principal loss quickly.

Liquidity / exit capacity

TVL scale, recent TVL change, and historical coverage.

Examples: Thin pools, sharp TVL drawdowns, missing TVL observations.

Why it matters: Exit capacity affects whether you can leave when conditions change.

Rate durability

Reward share, current rate vs median, volatility, and promotional end dates.

Examples: Incentive-heavy rates, spikes above 30-day median, promotional tiers.

Why it matters: Headline APY may not persist if incentives or demand shift.

Exposure / strategy complexity

Product type complexity, impermanent loss, leverage, and restaking layers.

Examples: LP impermanent loss, vault strategies, restaking slashing exposure.

Why it matters: Complex strategies stack risks that headline APY does not show.

Custody / counterparty

Operator identity, withdrawal terms, proof of reserves, and rehypothecation disclosure.

Examples: Centralized earn products, unclear withdrawal controls.

Why it matters: Custodial products introduce platform risk not present in noncustodial markets.

Terms / redemption

Lockup, unbonding, withdrawal queues, and whether material terms are known.

Examples: 30-day lockup, 14-day unbonding, unknown redemption terms.

Why it matters: Exit friction can extend beyond the APY horizon you are comparing.

Interconnection / dependencies

External protocol dependencies, bridge reliance, and restaking layers.

Examples: Cross-chain bridges, stacked protocol dependencies.

Why it matters: Failures in dependent systems can affect products that appear standalone.

Data quality

Field lineage, asset identity confidence, history coverage, and freshness.

Examples: Missing lineage, symbol-only identity, stale observations.

Why it matters: Risk and return summaries are only as reliable as underlying observations.

Risk dimension states

Lower concern

Evidence suggests relatively lower concern within this dimension.

Moderate concern

Evidence reveals meaningful considerations without implying a probability of loss.

Higher concern

Evidence reveals material concerns within this dimension.

Unknown

Yield.ly uses Unknown when material evidence cannot be verified instead of assuming a neutral value.

Not applicable

Dimension does not apply to this product type.

Observed-risk bands

Lower observed risk

Relative to Yield.ly methodology across available dimensions. Not zero risk.

Moderate observed risk

Meaningful evidence of concern in one or more dimensions.

Higher observed risk

Stronger evidence of concern across dimensions.

Insufficient data

Not enough verified evidence to classify confidently.

Evidence confidence

Risk concern and evidence confidence are separate.

High

Strong coverage of applicable indicators with verified sources.

Medium

Partial coverage with some verified evidence.

Low

Limited coverage; conclusions rely on sparse evidence.

Insufficient

Not enough evidence to support a meaningful classification.

Moderate observed risk with high confidence can coexist when strong evidence reveals meaningful concerns.

Unknown as a trust feature

Yield.ly uses Unknown when material evidence cannot be verified instead of assuming a neutral value.

Oracle marked Unknown does not mean no oracle risk. It means Yield.ly could not verify enough information.

Risk source hierarchy

1. On-chain or verifiable state
2. Official protocol documentation
3. Published audits and security reports
4. Structured provider data
5. Independent research
6. Curated internal review

Important limitations

• The presence of an audit does not mean a protocol cannot fail.
• The absence of a known incident does not prove safety.
• Yield.ly does not treat stablecoin as a synonym for safe.
• Higher TVL can indicate greater liquidity or adoption, but it does not prove safety.
• Assets held by a platform introduce custody and counterparty risk that does not exist in the same way for noncustodial products.
• Leveraged or recursive strategies can introduce liquidation risk and amplify losses.
• LP positions can experience impermanent loss and exposure to multiple assets. High fee APY does not eliminate that exposure.

Risk methodology version

Risk methodology version: risk@1.0.0. Last updated 2026-08-25.

• 2026-08-24: Initial observed-risk methodology v1.0.0 with eleven evidence dimensions.