A share is two things.
Buy the one you wanted. Drag the coupon sheet away from the certificate: what stays is a claim on the price, what comes away is a claim on the coupons. The perforation is drawn from this chain — an open hole is a week something was actually paid.
Implied annual coupon
2.46%
$100.99 × 2.46% × 0.334 years = $0.83 of coupon, leaving $100.16 of price. Undiscounted: there is no rate curve on this chain to discount against, and inventing one would make the number look better than it is.
Observed on chain
2.46%
2 coupons observed. The complete period 2026-07-08 to 2026-08-07 accreted 0.2022%, which annualises to 2.46%.
2 of 10 weekly periods since this chain’s first corporate action paid a coupon on SGOV. Perforation read 2026-08-31. The sheet rests where the chain left it; drag it to price a different coupon. Nothing is deployed — this is a simulation of an instrument that does not exist yet, priced from coupons that do.
What you are buying
Nobody wants both halves in the ratio the market hands them.
Buy a share and you buy two different assets welded together: a claim on where the price goes, and a claim on whatever the company pays out. Someone who came for the payouts is forced to carry price risk they never wanted. Someone who came for the growth is paying for a payout stream they would rather not fund.
Separating the two is four hundred years old. It is called STRIPS in Treasuries, and it has always needed a custodian, a paying agent and a legal wrapper — which is why it has always been for institutions. On this chain the payout is not cash landing in a brokerage account. It is a number on a contract. The custodian becomes a subtraction, and the apparatus collapses into a vault and two ordinary ERC-20s.
PRICE — the corpus
Captures everything the share price does. Redeems 1:1 for the underlying after maturity, with no deadline, ever.
It captures no coupons. If the company triples its dividend, PRICE holders see none of it.
KOUPON — the coupon sheet
Captures the accretions the token’s multiplier reflects, up to the last coupon. After maturity it stops accruing and is worth its undrawn share, then nothing.
It captures no price. If the share doubles, KOUPON holders see none of it.
KOUPON captures dividends if and when they are declared and reflected in the token's multiplier. That is not the same thing as a shareholder's dividend. A company can cut its dividend to zero, and KOUPON goes to zero with it.
Read from chain just now
The number this all rests on.
A dividend on this chain is not a payment. It is a movement in uiMultiplier() on the token contract. So the whole protocol reduces to reading one number correctly, and refusing to guess when it cannot be read.
| Token | uiMultiplier() | Contract |
|---|---|---|
| SGOV | — | 0x92FD…F9B5 |
| AAPL | — | 0xaF3D…93f9 |
| CRWD | — | 0xea72…3931 |
| NVDA | — | 0xd060…9EEC |
The chain did not answer, so every value above is a dash. Koupon never substitutes 1.0 for a number it could not read. All 194 tokens →
The mechanism
How a coupon gets from the chain to a holder.
1
The multiplier moves
The vault holds raw tokens and tracks the multiplier at the moment of each deposit. Accretion is computed from observed state transitions, never by summing events — the same corporate action has been emitted twice on this chain, and an event-summing accumulator would double-count it.
2
It is classified, and often held
The move is sorted by size into coupon, ambiguous or capital. Only the outer bands resolve on their own. Accrual keys to effectiveAt, not to the block timestamp — one action here was emitted three days before it took effect.
3
It is swept during market hours
A released accretion is sold to USDG in batches, market hours only, with slippage capped per leg. Tokenised equities trade around the clock; the things they track do not, and spreads widen hard when the exchange is shut.
4
KOUPON holders draw against it
Proceeds land in a claim pool for that underlying and maturity, drawn pro rata. Every division truncates toward the vault, so the last redeemer is never short.
A multiplier that cannot be read reverts. It never falls back to 1.0, never to a cached value, never to a default, and in the interface it renders as an em dash. A multiplier that moves down halts the sweep entirely: that has happened here once already, and a vault that has already paid out a reversed accretion is short.
The dangerous part
Splits are not coupons, and the chain will not tell you which is which.
A 2-for-1 split doubles the multiplier. So does nothing else, ever, that a coupon holder is entitled to. If Koupon read a split as a coupon it would hand KOUPON holders half the company.
So you would want the contract to say which it was. It does not. updateMultiplier() carries no reason code, and UIMultiplierUpdated(old, new, effectiveAt) has no action type. Three numbers, and none of them is the answer. The only signal available anywhere on chain is how big the move was.
Coupon
1.0 < r ≤ 1.005
Dividend-shaped. Observed automatically, released after a timelock and a second source.
Ambiguous
1.005 < r < 1.20
Never auto-releases. Held, published, and paid only when an external corporate-action feed confirms an ex-date and a matching size.
Capital
r ≥ 1.20
Split-shaped. Credited to PRICE, never to KOUPON. No sweep.
The case in the middle, 2026-08-31
Carnival’s multiplier moved +2.1486% — ten times larger than the biggest coupon this chain has ever paid, and fifty times smaller than its smallest split. On a $24 stock it is too big for an ordinary quarterly dividend and far too small to be a split.
I cannot tell you what it was from chain data, and neither can a contract. A threshold at 1% calls it a split and pays coupon holders nothing. A threshold at 5% calls it a coupon and, if it was a small stock dividend, overpays them. Both are guesses.
That is why the middle band never releases on its own, and why Koupon depends on an external corporate-action feed to pay anything at all. It is a trust assumption, it is not decentralised, and it is on this page rather than in the docs. See the transaction →
And it is not hypothetical that one underlying does both. WEEK paid a +0.3091% coupon on 2026-06-24 and split +100% five days later. The classifier does not get to specialise.
What could kill this
The coupons are thin, and the deep names do not pay them.
Stripping a coupon that rounds to nothing is a party trick, not a product. So here is every eligible underlying on this chain, its entire tokenised float, and every coupon it has actually paid — read on 2026-08-31, not estimated.
The problem is not that the numbers are small. It is that they are in the wrong places. NVDA has $10.5m of float and has never paid a coupon here. SCHD — a dividend ETF, the single most obvious thing to strip in this entire universe — has $316 of float and eleven holders. Depth and coupon are anti-correlated on this chain, and there is exactly one exception.
That exception is SGOV, a T-bill ETF: $713,004 of float, 414 holders, and the only token on this chain that has paid more than once. It is the reason Koupon has a first market at all, and it means its first honest market is a Treasury product rather than an equity one. If that table thins out, so does this protocol, and you should be able to see it before you use it.
| Underlying | Tokenised float | Holders | Coupons paid | Total accretion | Contract |
|---|---|---|---|---|---|
| SPY | $10,866,815 | 43,481 | none | — | 0x117c…4C0C |
| NVDA | $10,559,182 | 84,289 | none | — | 0xd060…9EEC |
| AAPL | $3,027,197 | 54,693 | 1 | +0.0566% | 0xaF3D…93f9 |
| MU | $2,068,702 | 29,702 | 1 | +0.0075% | 0xfF08…4afD |
| MSFT | $1,285,013 | 39,593 | none | — | 0xe932…2e74 |
| SGOV | $713,004 | 414 | 2 | +0.2982% | 0x92FD…F9B5 |
| ORCL | $515,257 | 18,872 | 1 | +0.2211% | 0xb099…EE03 |
| COST | $506,667 | 10,216 | 1 | +0.0612% | 0x4EA0…44C2 |
| DELL | $50,300 | 2,658 | 1 | +0.0064% | 0x941A…11Dd |
| IBM | $50,112 | 58 | none | — | 0x980d…9619 |
| ASML | $26,296 | 176 | 1 | +0.0101% | 0x47F9…dAEA |
| XOM | $12,907 | 2,863 | none | — | 0xf9B4…a2d5 |
| CSCO | $3,901 | 33 | none | — | 0xF543…a0dA |
| CCL | $3,733 | 41 | 1 unclassified | (+2.1486%) | 0x9651…aef9 |
| JNJ | $2,860 | 52 | none | — | 0x03Df…8c80 |
| VTI | $1,952 | 14 | none | — | 0x0594…6D09 |
| PFE | $1,664 | 10 | none | — | 0x7066…F51e |
| LMT | $1,522 | 20 | none | — | 0x329f…043c |
| UPS | $638 | 41 | none | — | 0xf232…9CE2 |
| SHY | $395 | 10 | none | — | 0xBE27…6FD8 |
| SCHD | $316 | 11 | none | — | 0xd63A…3C1D |
CCL is the row to look at. It shows 1 unclassified and its accretion is in brackets, because a +2.1486% move is not nothing and is not yet a coupon. Counting it as a coupon would overstate what Koupon can pay; counting it as “none” would hide a real corporate action. It sits in the ambiguous band until something off-chain says what it was.
Only one number on this site is annualised: SGOV’s 2.46%, from the one complete accrual period this chain has produced (2026-07-08 to 2026-08-07, +0.2022%, by day count). Everything else has been observed once or never, and one observation cannot be annualised without assuming a payment frequency the chain never states — so it is not. There are no gross dividend yields here and nothing has been grossed up.
And the one no design survives
The contract every Stock Token runs exposes adminBurn(address, uint256), pause() and pauseOracle(). The issuer can burn tokens out of any address — including a Koupon vault — and can freeze transfers and the multiplier.
Being non-custodial does not help. Being permissionless does not help. If Robinhood burns the collateral, the vault is empty and both legs are worth nothing. That is a larger risk to a depositor than anything in the mechanism, there is no architectural fix for it, and it would be dishonest to put it anywhere but here.
Getting out
Hold both halves and you have the share back.
Rejoining
Burn equal amounts of PRICE and KOUPON, take the underlying back. Any time before maturity, permissionless and fee-free. This is the arbitrage that keeps the two legs adding up to a share, and charging for it would break the thing it exists to defend.
The last coupon
At maturity KOUPON stops accruing. It stays worth its undrawn share of the claim pool, and once that is drawn it is worth nothing — which the market page says plainly for every underlying.
Redemption never expires
PRICE redeems 1:1 for the underlying after maturity, forever. No deadline, no expiry, and no sweep of unclaimed assets at any point, for any reason.
Four maturities per underlying, on a fixed quarterly ladder — Sep 2026, Dec 2026, Mar 2027, Jun 2027. Arbitrary maturities would fragment the liquidity into dust across a hundred markets, and a market only opens where there is a pool deep enough to sweep an accretion into.
If you want to know when it exists
There is nothing to use yet.
No contracts are deployed. The multiplier readings on this site are real; everything about stripping is a simulation, and says so wherever it appears. The waitlist is the only thing here that does anything.