Magnificent
Seven
$SEVEN is an ERC-20 that physically holds all seven of them (NVDA, MSFT, AAPL, GOOGL, AMZN, TSLA and SPCX) at equal weight. 7,777 machines compete for the stock the protocol buys with its earnings — and you have to burn $MAG7 to keep one running.
PFPs that
hold stock
Every token arrives holding $MAG7. Burn it to switch the token on, unlock a share on the skill tree, and take a cut of the stock the collection buys every single day. Pull the stock out whenever you like — the token keeps running.
A machine, loaded with fuel
The mint buys no stock. What it gives you is a token with $MAG7 sealed inside it, and $MAG7 is the only thing that switches a machine on.
Once a day, the pot is split
Royalties and token fees buy $SEVEN every day and divide it by weight across every machine still running. A machine that lapses stops counting.
One call, and you keep the NFT
Your stock leaves as seven ordinary tokens with no restrictions on them. The Genesis stays yours and starts earning again the same day.
7,777 of them, 26 one of one
The row above is ordinary tokens, drawn across the whole range - it is what a mint actually looks like. Every image is final and pinned to IPFS. The mint has not opened: price and date are set on deploy, and the mint page reads both from chain.
One holds the stocks.
One gets burned for them.
They do opposite jobs, and combining them into one token would break both. An index has to track its basket exactly; the fuel has to be free to be worth whatever people will pay to keep earning.
Its price is arithmetic
One share is a pro-rata claim on seven tokenized stocks the contract physically holds. Its price is the basket divided by the supply, and nothing else. Free to enter, 0.20% to leave, and that exit fee is the tracking error: it is exactly how far a share can drift from the basket before arbitrage closes the gap.
- Redeemable for USDG or for the seven stocks
- Enter with cash, or deliver the basket in kind
- 1%/yr management fee, the way an index fund charges
You have to burn it to earn
Fixed supply, no mint function, no peg. It is what a Genesis burns to switch itself on for thirty days, and the rent rises every period. Nothing is bought back and nothing is redeemed — the supply simply falls every time anyone keeps a machine running, and what it is worth is whatever people will pay for the right to keep earning.
- Burned on activation, and on every renewal after
- Airdropped inside each Genesis at mint, from a fixed reserve
- Supply and per-token allocation are not decided yet
The premium is the point
When revenue buys $SEVEN on the market rather than minting it, the buying pressure lifts it a hair above what the basket is worth. That gap is an open invitation: anyone holding tokenized NVDA or MSFT can deliver the basket, receive $SEVEN at exactly its net asset value, and sell into that bid. The protocol stops competing for the thin stock pools and starts paying other people to bring stock in. It only takes a fraction of a percent, which is why the index's fees are kept small.
The index is live and its contracts are verifiable today. The protocol documentation covers what exists. $MAG7 is written and tested but not deployed, and nothing here is a commitment to a date.
A share is the reward
for finishing, not for arriving
A freshly minted token earns nothing at all. It cannot even be switched on until it has unlocked something, and money cannot buy the first step. That is deliberate: a pot split by how many tokens exist rewards nobody, and a pot split by who did the work rewards the people who showed up.
17 nodes, four disciplines
Operator is what the contract can watch you do. Capital is read straight from your balances. Ritual is elapsed time, and is the one branch money cannot shorten. Signal is the social half — posting, your picture, bringing people — and because no contract can check those, they are signed by a named signer whose exact powers are stated on the page rather than buried.
The whole tree is worth exactly 1.00x and the contract refuses any change that would push it higher, so adding quests later cannot dilute anyone who already finished.
Escalating, on purpose
Every activation burns $MAG7 and lasts thirty days. The minimum rises each period, so keeping a machine running costs more every time — and the collection's daily income is divided between whoever is still willing to pay it.
That is the uncomfortable half, said plainly: this is designed to price people out. A machine that lapses can be swept by anyone and stops counting, and the share it was taking goes to everyone still running. Nothing is confiscated — the stock it already earned stays claimable, and the $MAG7 still inside it stays withdrawable.
Above 1.00x, burn buys weight on a square root to a ceiling of 10.00x. A hundred times the burn buys ten times the boost, so the tenth share costs roughly a hundred times what the second did.
Every stream buys the same thing
Two revenues, one direction, and a supply that only falls. Nothing routes to a wallet on the way past, and turning the crank is permissionless — a flywheel that only spins when the team pays gas is not a flywheel.
- 01
Revenue arrives
A 6.66% royalty on every secondary sale, enforced, plus a 5% fee on every $MAG7 sale. Both land as USDG.
- 02
70% buys $SEVEN
Which means it buys the seven tokenized stocks on-chain. The other 30% funds the protocol, and the split is fixed with no setter to lower it.
- 03
It divides by weight, daily
Only machines that are switched on count. Weight is what the skill tree unlocked, plus whatever burn bought above it.
- 04
Rent rises, claimants fall
Activation burns $MAG7 permanently and lasts thirty days, and the rent escalates every period. The same pot keeps being shared between fewer machines.
Held.
Not promised.
It owns the stocks
Not a synthetic, not a perp, not a price feed in a wrapper. Seven real balances in one contract, and you can read every one.
Equal weight, because it is checkable
Cap weighting needs a share count no contract can verify. Equal weight needs only balances and feeds. 14.28% per leg.
Two doors out, always
Redeem for USDG, or take the seven stocks directly. The in-kind exit reads no oracle and calls no router, so it survives both being broken.
Two doors in, as well
Send USDG and it buys the basket, or deliver the basket yourself. The second touches no pool, so its size is bounded by what stock exists rather than by what pools will sell.
The protocol owns its own liquidity
Every fee the pool earns comes back to the collection instead of to whoever happened to supply it.
An unauthorised pool can never be funded
Creating one is not preventable, and we do not pretend otherwise: it never touches the token. But both ends of every transfer must be a wallet or an approved contract, so a rogue pool stays an empty shell.
Swapping open, provision closed
A contract moving somebody else's shares needs approving separately. A swap router can be cleared without a position manager being able to add liquidity beside us.
Locked principal, open fees
The position is held directly, so there is no NFT to move. Principal sits behind a deadline that only moves forward; collecting fees is permissionless and never gated.
The exit fee is the tracking error
Arbitrage below NAV costs exactly the exit fee, so that fee is how far a share may drift from the basket. At 0.20% the band is about a quarter of a percent.
No hidden edge
A transfer tax is not an option at any size. On a v3 venue a sell transfers into the pool, the pool receives less than it requires and reverts, while a buy transfers out and succeeds. Buys working and sells reverting is the shape of a scam whether or not anyone meant it, so the fee lives on redemption instead. 80% of what it collects goes to the collection's backing and the rest to the treasury. All of it is itemised in the docs.
What stops it going wrong, and what does not
The exit is never the owner's to close
Minting can be paused. Redemption has no switch: not paused, not timelocked, not at anyone's discretion.
The basket cannot be admin-drained
The rescue hatch is announced three days ahead and reverts on any constituent. It reaches strays and dust, nothing else.
Stale prices halt the vault
Stock feeds sleep nights and weekends. Past six hours without a print, mint and redeem both refuse rather than trade against a frozen reference.
Rebalancing is capped and rate-limited
Only past ±20% of target, at most $200 per leg, then that leg freezes for a day. Thin pools are not worth poking.
This is a permissioned token
Multisigs, bridges and exchange deposit addresses are contracts, and each needs approving before it can hold shares. That is the cost of owning the market.
The periphery can be changed
The collection and backing sit behind proxies the owner can upgrade, so their guarantees are 'will not' rather than 'cannot'. The vault holding the assets has no upgrade path at all.
The array is online
Live NAV, per-leg balances, weight drift and feed freshness, all read straight from chain.