Methodology

Most DeFi rates come from the DefiLlama yields API, matched to our own list of venues by pool ID rather than by token symbol. Symbols collide across chains and issuers — there is a SPY on Robinhood Chain and a wSPYx on Ethereum — so a symbol match would quietly mix up two different products.

Aave V4 and Morpho come from their own GraphQL APIs, which cover markets DefiLlama does not index, and Kraken’s vaults are read directly from Ink, Kraken’s own L2. Anything no API covers is entered by hand and labelled as such, and rows sourced by hand carry the date a human last checked them. A single row can draw on more than one source: Kraken’s vaults take their size from the chain and their rate from a human reading Kraken’s page, because no one source has both.

defillama

28 rows

aave_api

7 rows

onchain

3 rows

morpho_api

12 rows

manual

11 rows

A scheduled job runs every 60 minutes and writes a snapshot. The last successful run was 23 Oct 2026. Every snapshot is committed to version control, so the history of any number on this site can be reconstructed.

Each market links to the most specific page we can establish for it, which for most DeFi venues is the market or lending page rather than the protocol’s front door. The arrow marks an outbound link; hovering says whether it is market-specific.

These are resolved from each protocol’s own identifiers, not guessed: Kamino market and reserve addresses and Jupiter vault ids from their APIs, Morpho market ids cross-checked against the markets listed in their sitemap, and Aave reserves by symbol. Where two markets shared a collateral, loan asset and LLTV, we used on-chain activity to tell them apart rather than picking one.

Crypto.com and Kraken have no public per-market URL, so those rows link to the product page. Every other market links to its own vault, reserve or market page.

The timestamp in the last column links to that pool’s rate and TVL history on DefiLlama, where we read it from. That link is always exact, because the pool id is the one we query.

Aave V4 reserves come from Aave’s own GraphQL API, which DefiLlama does not index for the equity spokes. It gives the collateral factor, the liquidation bonus and the borrow rate on the loan asset directly from the protocol. Everything else comes from DefiLlama’s lendBorrow endpoint, read from the collateral market itself. The borrow cost is the rate on the asset you would actually borrow, which is not always the same pool: in a pooled market such as Kamino’s xStocks Market, the stock reserve’s own borrow rate is the cost of borrowing the stock, so we point at the stablecoin reserve in that market instead and record which pool each rate came from.

We do not publish available borrowing liquidity. Supply, borrowed amounts and debt ceilings mean different things in a pooled market and an isolated one, and a single “available” column would imply a comparability that is not there.

Base APY is interest paid by borrowers. Reward APY is a token incentive the protocol is paying out of its own budget; it can be cut at any time. Points are a promise of a future airdrop with no stated value. We show all three separately and never fold points into an APY figure, because nobody can honestly price them.

Incentive programmes stop. When one does, we mark the opportunity Ended rather than deleting it: the row keeps the last rate we recorded, shown struck through, with the date and what we observed. Ended rows are excluded from every headline figure — the median, the best rate, the totals and the borrow tables — so a programme that no longer pays cannot flatter the numbers.

We keep them because a venue that paid for a while and then stopped is the most direct evidence this site has for its own argument. A yield aggregator that quietly removes the rows loses that history.

Anyone can create a Morpho Blue market, and Morpho only surfaces curated ones in its own interface. Several of the largest tokenized-stock markets are not listed there: one NVDA market holds about $1.4m of collateral and pays a 39% borrow rate, and you will not find it by browsing. It is reachable by direct link and the collateral is real, so we list it — flagged Unlisted, because an uncurated market’s oracle and parameters have had nobody’s eyes on them.

Where several markets exist for the same stock and loan asset, we show each as its own row rather than picking one. They are genuinely different markets with different rates, sizes and risk.

Morpho lists a “Lend Rate” against markets whose collateral is a tokenized stock, and on some of them it is very high — over 50% on one Robinhood Chain market as we write. It is easy to read that as yield on the stock. It is not. Morpho’s documentation is explicit that lenders deposit the loan asset and that collateral is posted by borrowers and never lent out, so that rate belongs to whoever supplies the USDG or USDC.

We show it in the borrow table as “Lend side”, clearly separated from what the stock earns, which on Morpho is zero. Folding it into the stock’s APY would make this site exactly as misleading as the pages it is meant to clarify.

Several venues accept tokenized stocks as collateral but never lend them out. Those rows show 0% because zero is the correct answer. Rows where we simply have not collected a rate say not collected instead. These are different things and the table never blurs them.

• Marketing ceilings. “Up to 4%” is not a rate and does not go in the table.
• Estimated or back-filled APYs. If we did not read it from a source, it is null.
• A rate without a timestamp. Every row shows when it was last collected, and rows older than twice the refresh interval are flagged as stale.

• We do not publish liquidation thresholds as a column, only the maximum loan-to-value. Aave reports a 5.5% liquidation bonus on its equity reserves; other venues express the same risk differently, and a single column would flatten distinctions that matter.
• Superform’s own app and DefiLlama disagree on its vault rates by roughly a factor of 1.8 — on 19 Oct 2026 Superform showed 12.56% on superTSLA where DefiLlama reported 22.98%, with both reporting the same deposits. The rate is sUP emissions in both cases, so the gap is how that token is valued and annualised. We publish DefiLlama’s figure because it is the source we can read hourly, which means our highest headline rate is the higher of two readings. Check the vault page before trusting it.
• Crypto.com publishes a per-stock rate table but does not say whether the yield is interest, an incentive or revenue share. We record the total and leave the split null, and those rows are excluded from the “yield that is real interest” figure rather than counted as incentives.
• Bybit and Bitrue both list tokenized stocks for spot trading but pay nothing on them. They stay on the watchlist so the absence is documented rather than implied.
• Kraken’s vault sizes are read from the Veda BoringVault contracts on Ink, valued at the vault’s share supply times the price of the underlying xStock. The exact share rate lives in an Accountant contract whose address we could not find, so this understates each vault by roughly the yield it has accrued since launch — on the order of a percent. The rates themselves are hand-entered from Kraken’s site, because their Earn API needs an authenticated key.
• CeFi has no usable public API for tokenized stock earn products. Kraken’s vault rates need an on-chain reader we have not built yet; Bitrue and Crypto.com need manual checks.
• DefiLlama reports no reward APY for the Kamino xStocks markets even though incentive programmes have been announced there, so our reward column may understate them.
• TVL on CeFi venues is almost never disclosed. Those cells read “Not disclosed” rather than being left blank.
• 7 venues sit on the watchlist because we could not confirm a live rate.

The full dataset is available as JSON at /api/opportunities and as CSV at /api/opportunities.csv. It is free to use. A link back is appreciated but not required.

If a number here is wrong, it is worth telling us — the fix usually takes one line in a seed file. Every opportunity carries a verified flag that only a human can set.