kerne-protocol/solana-yield-honesty

Solana Honesty Index

What each Solana stablecoin product says it pays, next to what it actually paid, measured from a share price rather than from a claim.

Snapshot generated 2026-09-22T12:05:13.833Z. Window 30 days. 13 products across 3 protocols, 13 comparable, 0 published but not comparable. Realized figures: 5 by issuersharepricehistory, 2 by onchainshareprice, 6 by issuersharepriceobserved.

Widest gap among rows this board can rank: kamino-lend-pyusd, at 0.18 percent against the median of its own advertised history.

Not every row on this board is denominated in dollars. Jupiter Lend Earn EURC (EURC). Each such row is internally consistent, because its advertised and realized figures are both measured in its own unit, so its gap_pct is meaningful. Sorting the advertised_pct column across the whole board is not, because it puts two currencies in one ranking. The symbol column is what tells them apart.

Read the gap column with this next to it

A lending reserve's advertised supply APY is an instantaneous rate set by utilisation, not a forecast and not a trailing average. On a reserve sitting near a kink in its own rate curve, that number moves by multiples within a single day while the realized 30 day figure moves by hundredths of a point. Subtracting a 30 day realized figure from one spot reading therefore measures when the collector ran at least as much as it measures the product.

This is not hypothetical and it is not somebody else's mistake. This project reported a 2.32 point gap on Kamino's PYUSD reserve on 2026-08-04 and a gap of the opposite sign on the same reserve three days later, and very nearly published the first one as a finding. So every row that can support it now carries the distribution its spot reading was drawn from:

Widest range in this snapshot: jupiter-lend-usdg, whose advertised figure ran from 0.56 percent to 51.30 percent over the window, a factor of 91.34.

5 row(s) in this snapshot were captured outside the middle half of their own recent range (kamino-lend-usdg, kamino-lend-pyusd, jupiter-lend-jupusd, jupiter-lend-usdt, jupiter-lend-usdg). For those rows, prefer gap_vs_median_pct over gap_pct.

gap_vs_median_pct is the same subtraction done against the median of the advertised figure's own published history rather than against one reading of it. Where both exist, it is the more honest number, and it is the one to quote.

If one of these products is yours

Everything above is free, and stays free whether or not anybody ever buys anything. Nothing on this board is for sale: no row moves for money, no row is added or removed for money, and no protocol here has been contacted about its row. That is the boundary, and this section is the only commercial passage in this file.

There is a longer version of this measurement that a protocol can commission about itself: <https://kerne.fi/disclosure-audit?src=hf-sol>

The products below are picked out of this snapshot, closest to exact delivery first, rather than typed into this file. Today the rows landing nearest their own advertised rate are Kamino Lend USDS (Main market) (98.74 percent of what it advertises) and Jupiter Lend Earn USDT (96.83 percent of what it advertises).

What it is: a Disclosure Integrity Audit, 499 US dollars flat, one report inside 72 hours of scope confirmation, commissioned by the protocol being reviewed and delivered privately to it. It reviews whether your public claims match your chain on three axes: advertised against realized yield (the method in this dataset, run properly over your whole surface rather than one reserve), the addresses and figures in your documentation against the live registry, and your oracle and attestation posture. It ends in a signed findings summary you are free to publish or to never mention again.

What it is not: it is disclosure review, not a security audit and not assurance. It certifies nothing about your security, your solvency or your compliance, and it is not a rating. It is also not this board: commissioning one does not add, move or remove a row here.

How realized is measured

A share price is the only measure of a yield product a holder cannot be talked out of. It moves when value lands and it does not move when value does not.

Solana has no ERC-4626, so there is no single call that returns a share price across products. Each adapter is responsible for producing a series of { timestampSeconds, pricePerShare } and everything after that is one piece of arithmetic applied identically to every row:

annualized = ((ppsTo / ppsFrom) ^ (365 / days) - 1) * 100

The realized_method column says how each row's series was obtained, and the three values are not equally strong:

• —onchain_share_price reads the protocol's own account data over plain JSON-RPC and derives the share price ourselves.
• —issuer_share_price_history uses a share price series the issuer publishes.
• —issuer_share_price_observed is a share price this collector samples once per run and accumulates itself. The issuer reports the assets; the share count is the SPL mint supply, which this project reads on chain and holds against the reported figure on every run. Half independently verified, half not.
• —thirdparty_rate_series is the weakest, a time weighted mean of a third party's daily rate observations, used only where no share price series exists yet. A row on this method has not been independently measured, and it says so.

Where a figure was converted rather than quoted

Not every issuer publishes in the units this board compares in. Jupiter quotes supplyRate as a simple annual rate; the realized figure here is a compounded annualization of a share price. Placing the two side by side untouched would manufacture a gap out of pure arithmetic, so the advertised figure is converted to its daily compounded equivalent before comparison.

Any row where that happened says so in advertised_transform, and advertised_verbatim still carries the issuer's raw payload so the conversion can be checked or rejected. An empty advertised_transform means the number in advertised_pct is the issuer's own, unaltered.

The comparability gate

A protocol quoting a since inception average is not answering the same question as a protocol quoting what a holder earns today. Ranking a board by whichever number is largest on each protocol's own page produces a ranking of marketing copy, so this dataset refuses to do it.

An advertised figure is compared only when its basis is spot-apy or trailing-30d. Everything else is still published, with the reason it was not compared in the reason column. Where a protocol publishes several comparable figures, the most conservative one is used, never the largest, so no protocol can improve its standing here by adding a bigger number to its page.

Every row in this snapshot was comparable.

Columns

Why the publisher is not on this board

This dataset is built by Kerne, and there is no Kerne row on it. That is the first thing a reader should be suspicious of, so here is the reason and the place to check it rather than an omission left to be noticed.

This board measures Solana products. Kerne's products are on Base. There is no Kerne row to add here that would not be invented, and inventing one to look even handed would be the same failure this dataset exists to measure.

Kerne does publish itself, by the same method, on the EVM board at huggingface.co/datasets/kerne-protocol/honesty-index, where its own row is flagged is_kerne and is measured by the same share price arithmetic as every other protocol on it. It does not come out of that well. Read the row rather than this paragraph: it refreshes daily and this paragraph does not, and a number typed here would be a false claim the moment it moved.

If Kerne ever ships a Solana product it goes on this board, measured by this code, with no exemption. Until then, absence here is a scope boundary and not a favour.

Licence and scope

The compilation and every column computed here are released under CC BY 4.0.

advertised_verbatim is short factual quotation from each issuer's own public surface, attributed and dated, and is not claimed under that licence. Protocol names and marks are used nominatively to identify the products measured.

Limitations, stated rather than buried

• —13 of 13 rows in this snapshot carry a comparable pair. The rest are published with their reason.
• —Rows on thirdparty_rate_series depend on a third party and are labelled as such.
• —The independently collected share price series in this repository starts on the day this collector first ran. It does not backfill, and a row will say insufficient_history rather than annualize a window it does not have.
• —A figure read from an issuer API is the figure that API served at captured_at. It is not a claim about what a human sees in that issuer's app at any other moment.