Data & methodology

Every rate Lendwise shows is produced by the same pipeline, so numbers are comparable across protocols and chains. This page documents exactly how.

Sources

For each protocol, Lendwise reads from the most authoritative source available and reconciles on-chain and off-chain data:

• Aave V3 — protocol subgraphs + on-chain reserve data.
• Morpho — Blue markets and MetaMorpho vaults via the Morpho API + on-chain.
• Compound V3 — per-chain subgraphs (schemas differ by chain) + on-chain.

When multiple sources are aggregated, one source failing never blocks the others — partial data is better than no data.

Cadence

Every 10 minutes   → collect a spot APY snapshot for every market → apy_hourly
Daily (00:10 UTC)  → aggregate the day's hourly rows → apy_daily

• Hourly rows are a running average per market per hour — resilient to a single noisy read.
• Daily rows are a single averaged value per market per day, and carry a completeness score: how many of the day's expected hourly slots were actually captured. Days below the reliability threshold are flagged and excluded from comparisons.

APR → APY

Protocols report rewards as APR. Lendwise converts every APR to APY with daily compounding before it's ever stored or compared:

APY = (1 + APR / 365) ^ 365 − 1

This matters: a raw 20% reward APR is ~22.1% APY. Comparing an APR from one protocol to an APY from another silently mis-ranks markets. Lendwise never does this.

Net APY

All rates are stored and compared as net APY, direction-aware:

• Supply net = base − fees + rewards
• Borrow net = base + fees − rewards

Each rate also keeps its full breakdown — base, total rewards, fees, and every individual reward token (with source: protocol, Merkl, or Merit) — so you can audit any headline number.

Normalization rules

A few rules keep cross-protocol data honest:

• Chains are identified by chain ID, never by name. Adapters spell the same chain differently ("Ethereum" vs "ethereum"); only the numeric chain ID is canonical.
• Product identity is structured, not parsed. Every market has typed fields — provider, chain, asset, kind (supply/borrow) — rather than a string that gets sliced. Filters hit indexed columns, not substrings.
• Rates are always APY. No mixing of APR and APY anywhere in the dataset.

Known limitations

• Rates are snapshots on a 10-minute cadence, not real-time streaming.
• A market appears only once it's been indexed; brand-new markets may lag by up to a cycle.
• Occasional discrepancies between a protocol's official API and its own UI are resolved in favor of the official API. Documented cases are tracked publicly.

Query it yourself

Everything above is exposed through the GraphQL API — hourly and daily APY, the full breakdown, market state (TVL, utilization, price), and product metadata.