2026-09-19 · Tokenized assets & RWA

StonkFun and STONK — "stock dividends" paid by the holders themselves

Data as of 19 Sep 2026.

Bottom line: a project with verifiable revenue, rare for its class — and four reasons not to rush. StonkFun is a venue where memecoins launch not against the dollar but against a tokenised stock — an S&P 500 tracker, say — and memecoin holders receive "dividends" in that same stock. In reality those dividends come from a tax on every transfer, paid by other traders of the same memecoin; nothing enters the system from outside. Venue revenue is readable on-chain, the burn checks out, mint authority is revoked — but the economics are 13 days old, Terms of Service don't exist, 40% of the flow goes to an undisclosed recipient, and the deflation that made the token attractive can't be repeated at the same price.

What it is

StonkFun is a memecoin launchpad on Solana. It has no on-chain program of its own: it's a configuration on top of Raydium LaunchLab, whose contracts are audited and governed by a multisig with a 24-hour timelock. Creators choose what their coin trades against: a tokenised stock (SPYx, NVDAx, TSLAx, MSFTx from Backed), a crypto asset (ZEC, HYPE, TAO), a commodity, or a synthetic share in a private company (OPENAI, ANTHROPIC from PreStocks). The venue token is STONK. Launches as of 19 Sep 2026: 59,971.

StonkFun daily fees, 26 Jul–18 Sep 2026. Grey — before the Raydium LaunchLab integration, red — after. Source: DefiLlama, series pulled 19 Sep 2026.

1. There is no collateral at all

The "dollar → stock → collateral → memecoin" chain describes an impression, not the mechanics:

• dollars become SPYx outside StonkFun — at Backed or on an exchange;
• on StonkFun the creator mints 1bn tokens from nothing — nothing is deposited or locked;
• the quote asset accumulates as pool reserves in Raydium program accounts — that's market depth, not backing, and it drains as people sell;
• there is no mechanism to redeem a memecoin for the underlying stock — a holder can only sell into the pool.

On STONK itself it shows in the numbers: liquidity against SPYx is $6.53m at a $200m market cap — 3.3%. That isn't "backed by a tokenised S&P 500".

2. "Stock dividends" are a transfer tax — and one key controls it

The venue's docs describe the mechanism plainly: holders pay a tax on every transfer from the first trade — 1% or 3%, fixed forever at launch (older launches take 85% of fees from their 4% pool). What's collected is distributed pro rata in whatever asset the pair is quoted against.

It isn't a dividend, for three separate reasons:

• The source isn't company profit but holders' own turnover. The same people pay and receive.
• Backed's tokenised stocks pay no cash dividends — the custodian receives the dividend and reinvests it, and the holder sees a rising multiplier. There's no real NVIDIA dividend to pass on.
• The distribution isn't trust-based. The right to withdraw withheld tax (withdrawWithheldAuthority) sits with a platform key, not a contract with guaranteed payout. The docs themselves allow that with the wrong configuration the tax accrues and is never distributed to anyone.

The machinery does work: one PEPE-quoted token has distributed 50.9bn units across 28,897 payouts to 3,476 holders. But the top distributions are crypto-quoted, not stocks.

3. What it costs a trader — the rate is known, the total isn't

The rates are published and verifiable: 1% on every trade (half to the creator, half to the venue; on the 2% tier the creator takes three quarters), plus 1% or 3% on every transfer for reward tokens.

The total taken from traders can't be stated. The denominator differs 9.3x: $280m of volume per DefiLlama, $1.86bn per the venue's own API, "over $2.6bn" per its own claim on 15 Sep. And the $1.86bn isn't a measurement — it's revenue divided by an assumed 1% rate. The later figure ($1.86bn, 19 Sep) is smaller than the earlier claim ($2.6bn, 15 Sep) — cumulative numbers don't behave like that.

Same with rewards: the claim grew from $5.35m (7 Sep) to "over $40m" (15 Sep) — 7.5x in eight days, with no distributor address or transaction hash. An independent third-party post puts the cumulative total at $3m. $40m would be double the venue's entire stated revenue and four times all buybacks: if it's real, money must be arriving from outside fees, and no such source shows up in the data.

4. The venue's whole economy is thirteen days

The median day jumped 23x overnight — on 6 September, when the venue moved to Raydium LaunchLab. The flywheel the project talks about has never been observed outside a single burst of attention.

The fair flip side: volume isn't fading. Peak on 11 Sep — $38.56m, low on 15 Sep — $14.18m (−63%), and by 18 Sep volume was back at $38.00m. Revenue recovered less (−37% from peak), and price fell 30% from its high. So price fell harder than venue activity. The "burst is already fading" thesis isn't confirmed yet — high volatility is.

5. The burn is real — but it can't be repeated at the same price

Measured on two independent Solana nodes (19 Sep 2026) and confirmed by two third-party mint reads (RugCheck, Jupiter) to the last digit:

The burn is the only project figure that survived reconciliation with the chain. Mint and freeze authority are both revoked: there is no button the issuer could use to confiscate tokens.

But here's what follows. Per the venue's own data, buybacks cost $9.68m and bought 161.5m tokens — an average of $0.0599, a quarter of the $0.238 price at the time of writing. Most tokens were bought in August, when STONK traded under a cent (low of $0.00175 on 5 August). The same supply reduction at current prices would cost $38.9m instead of $9.7m. A dollar of revenue now removes four times fewer tokens: past deflation shouldn't be extrapolated.

Two gaps in the dashboard's own accounting: 12.34m tokens bought but not marked as burned, and 10.69m burned from a source the dashboard doesn't itemise. Looks like a labelling artefact, but it can't be reconciled from outside.

6. 40% of the flow undisclosed, and no legal perimeter

The revenue page: 60% of revenue goes to buyback and burn, "The rest is retained". No address, no purpose, no recipient — for roughly $5.6–6.7m over two months. Alongside that:

• There are no Terms of Service. The text sits in every page footer but links nowhere; six URLs checked returned 404; there's no archive snapshot. No document covers jurisdiction, liability, restrictions for US persons or token status.
• The team is anonymous. No names, no entity, no jurisdiction. Domain registered 23 Jul 2026.
• No audit of the off-chain part. Raydium's contracts are audited, not what actually holds the money: tax collection, payout calculation and buyback execution are done by the venue's backend with its own key.

7. The rail underneath is the strongest link

The xStocks rail is stronger than the venue built on top of it — an unusual setup.

But there are two classes of quote asset, and the second is much worse. PreStocks (pairs against OPENAI, ANTHROPIC) holds stakes through a chain of SPVs; no Backed-level attestation was found. In May 2026 Anthropic and OpenAI publicly stated they don't permit SPVs to hold their shares and that such transfers are void — the tokens fell 30–40% at the time. In the UI these pairs sit next to SPYx and look identical.

8. A legal precedent aimed at the rewards mechanism

The US regulator's stance on memecoins has softened (statement of 27 Feb 2025: "akin to collectibles"), but it rests on one pillar — that value comes from hype, not from the efforts of others. Paying revenue to holders knocks that pillar out. In Terraform Labs (ruling of 28 Dec 2023) the court found the MIR token to be a security precisely because holders received 0.25% of trading fees. StonkFun's rewards mechanism is the same feature, only bigger. Alongside: in the case against pump.fun (the closest business analogue, same 1%-per-trade model), the court on 31 Aug 2026 allowed racketeering claims to proceed.

Scenarios

• A — attention fades (base case). Check on 30 September and 31 October: the 7-day average of daily fees. The "business outlived the burst" thesis survives only if on 31 Oct it's above $250k a day — a quarter of the mid-September level and seven times the median of the first 42 days.
• B — the quote asset disappears. Binance's tokenised stocks lasted about three months from launch to shutdown under regulatory pressure in 2021. Check on 31 October: SPYx versus SPY, and whether the next quarterly attestation came out.
• C — the rewards mechanism gets reclassified. No date: regulatory action has no schedule, but the feature has already triggered once in court.

Unknowns

Obtainable. Total venue revenue differs by 18–34% between the project's dashboard ($16.43m) and DefiLlama ($13.89m), and the gap is uneven day to day. Until someone runs their own on-chain query against the venue's program, the "60% of revenue to buyback" claim rests on a number nobody outside can see: the burn proves tokens were bought and destroyed, not that the money came from fees rather than treasury. Likewise, the share of tokens that "graduate" from the curve isn't established (13.1% per one dashboard, 2.6% per an independent sample), so the comparison with pump.fun is unprovable either way for now.

Unknowable from outside. Who the operator is. Where the 40% goes. Whether tax accrued on "non-adopted" launches will ever be distributed. How long launchpad revenue survives once the narrative fades — no example in this class has lived long enough to answer.

How much attention it deserves

Against "dismiss" — things this class almost never has: revenue readable on-chain by a third-party tool; a burn confirmed to 0.015%; revoked mint and freeze authority; moderate holder concentration; a quote-asset rail stronger than the venue.

Against "look now" — four things, each sufficient on its own: the economics are thirteen days old; there are no Terms of Service; 40% of the flow goes to an unknown recipient; the deflation cost a quarter of what it will cost from here. Most importantly: the 60/40 split is operator-wallet policy, not a smart-contract rule. A token whose cash flow depends 100% on memecoin trading and fails at least four of eight standard "tokenomics theatre" tests gets only a minimal slice of a portfolio under disciplined frameworks. A sensible review point is 31 Oct 2026: revenue and holder concentration.

Three ways to get exposure — and what each risks

Ranked by what it earns relative to what it risks.

1. Hold STONK — be the casino, not the player. 5/10. Traders on the venue pay 1% per trade and 1–3% per transfer; STONK sits on the other side of that flow. But a STONK holder receives nothing directly — no staking, no payouts; all the upside is supply reduction, which now costs four times as much. Add no Terms of Service and an undisclosed 40%.

2. Provide liquidity to the STONK/SPYx pool. 4/10, conditional. The main pool turns over about twice a day — that turnover is what pays LPs. But at this volatility, losses from price moves will almost certainly eat the fees, and neither leg of the pair is a dollar. Without measuring 30-day fee yield against σ²/8 over the same window, it's an assumption, not a strategy.

3. Trade the SPYx–SPY dislocation. 7/10 — the healthiest. The venue pushes real volume into thin pools of tokenised stocks whose true price is known to the cent — which creates dislocations. On a neighbouring rail (tokenised SPY), persistent weekend deviations of ±2% and a +4.44% spike have already been measured. The only serious link — the issuer — is solid (Clearstream, InCore, weekly Proof of Reserve, quarterly ISAE 3000). And the strategy doesn't need StonkFun to survive — it needs it to be noisy. First step: an hourly SPYx-vs-SPY series over two weeks, focused on hours when the exchange is closed.

What's not on the list. Launching a token to collect creator fees — that's the side taking money from retail. Shorting STONK — there's no perp, and shorting a token with a managed burn and thin float carries unlimited loss. And anything involving PreStocks pairs — a wrapper whose validity the underlying issuer has publicly denied isn't a strategy object from any angle.

Call

By 31 Oct 2026, StonkFun's 7-day average of daily fees won't exceed $250,000 a day. This is a bet on method: launchpad revenue that depends 100% on memecoin trading is fragile and forms no durable floor. If fees hold above a quarter of the peak seven weeks after the burst, it's that rule that gets refuted. Value on 18 Sep: $677,979 a day.

Sources: DefiLlama (fee and volume series via Dune, 19 Sep 2026) · DexScreener, CoinGecko, Jupiter (prices and pools) · Solana nodes (supply and mint authorities) · RugCheck, Jupiter · StonkFun docs and API (classed as project claims) · Bitquery (program addresses) · Backed documentation and reporting on the Kraken acquisition · court rulings in Terraform Labs and the pump.fun case.

Research for information only. Not investment advice.