Methodology
This page defines every figure VaultSight publishes. If a number is on the site, its formula, source, and known limitations are here.
Realized return
Computed from ERC-4626 share price (convertToAssets), never from a published APY — even for vaults where a published figure exists.
realized_return = (aps_end / aps_start) - 1 realized_apy = (aps_end / aps_start) ^ (365d / elapsed) - 1
In standard ERC-4626 accounting, fees dilute or accrue against share price, so realized return is net of vault fees by construction. This assumption is validated and documented per protocol.
Advertised rate
The time-weighted average of the published rate over the same window:
advertised_twa = Σ(published_apy_i × Δt_i) / Σ(Δt_i)
Every advertised figure is labeled with its source and the smoothing window it was computed over. Where multiple publishers are archived, all are shown — never merged. Unsourced advertised figures are never published.
The gap
gap_bps = (advertised_twa - realized_apy) × 10000 capture_ratio = realized_apy / advertised_twa
Capture ratio is the headline figure. It is undefined — not zero, not infinite — when advertised_twa <= 0, and is suppressed rather than displayed in that case.
Attribution
The gap alone is a provocation; the attribution is the product. Where the data supports it, a figure is decomposed into: annualization noise from short smoothing windows, rate decay after deposit, deposit dilution against capacity, and reward-token emissions present in the advertised figure but not in share price (flagged, not yet quantified). Fee timing is out of scope for this decomposition.
Windows and coverage
Standard reporting windows are 7d, 30d, and 90d, expanding as the archive grows. A window with insufficient archive coverage is shown as unavailable, never interpolated.
Every published figure carries a coverage percentage — the fraction of expected 5-minute observation slots actually present for that vault and window.
Independence
VaultSight is read-only and non-custodial, and is not affiliated with any vault protocol, curator, or aggregator it measures. Aggregators that publish their own rates (e.g. QuickNode Earn) are treated as a measured publisher, not as infrastructure — their capture ratio is a data point like any other vault's.
Revision log
No revisions yet. Methodology changes will be logged here with the date and what changed.