The fee flywheel

Investors Center funds itself from its own trading volume. The protocol owns the market for INVEST — five markets, in fact: one two-sided anchor and four one-sided sell walls — and each of them charges a fee. The fee is taken in whatever the trader is spending, converted to a stablecoin where it has to be, kept as-is where it is already something the treasury wants, and spent on a governed basket of reserve assets that are delivered straight into the Treasury as backing. Backing grows while supply does not — and that is what pays stakers.

INVEST itself is a plain ERC-20. Every transfer moves exactly what was sent, and there is no charge on holding it, sending it or receiving it. The revenue is a property of the pools, not of the token.

This page explains the mechanism and its guarantees. The holder-level version is The flywheel; the contracts that implement it are listed in Contracts.

The flywheel is stage 2, and stage 2 has not run

Everything below is written in the present tense and describes the finished mechanism. None of it is turning. Of the five pools, one vault is deployed and holds no liquidity and the other four do not exist; TaxCollector, FeeAllocator and ReserveAccumulator are not deployed at all; INVEST total supply is zero and the Treasury holds nothing. No fee has been charged, collected, split or spent. See Deployment status.

The loop

	Stage	What happens
I	Trade	Someone buys or sells INVEST on one of the five protocol-owned pools, and that pool takes 5% of what the trader is spending.
II	Collect	One permissionless call sweeps the accrued fees out of all five positions — INVEST from every sell, and each pool's own quote from its buys.
III	Convert, or keep	The INVEST from all five pools is sold for USDG — but only into the anchor, the one pool that holds stablecoin. The USDG fees need no sale. The stock fees are never sold at all: they go straight to the Treasury as the asset itself.
III b	Realise	When a sell wall has actually sold INVEST, the stock it was paid is position principal, not fee — invisible to stage II. A separate permissionless call hands it to the Treasury and re-sets that wall higher. This is the only route a wall's sale has to backing.
IV	Split	A decaying share goes to the team, in stablecoin and in kind; the rest goes to the accumulation desk.
V	Allocate	The desk fans the treasury's share across the destinations governance has approved, by weights the operator sets — with a floor under reserve accumulation.
VI	Accumulate	The accumulator buys the governed reserve basket and delivers it to the Treasury.

Everything delivered at stages III, III b and VI lands in the Treasury as pure backing. It is not a deposit against which INVEST is minted, so it adds reserves without adding supply.

I–II. The pool fee

The protocol's market for INVEST is five Uniswap V4 pools, each with terms pinned permanently: a 5.00% fee (50,000 pips, in the V4 fee unit), no hook, and a fixed price grid. Those things together are a pool's identity. Changing the fee would not re-price a pool; it would name a different one. There is no rate setter, no governance vote and no disable path.

Pool	Kind	Share of supply	Paired with	Why it exists
INVEST / USDG — the anchor	two-sided, full range, grid 32,767	60% (36,000 INVEST)	$10,000 of USDG	the stablecoin-quoted market, and the only pool that holds cash. It is the reference price, and the only pool the protocol ever sells INVEST into.
INVEST / RCAT	one-sided sell wall, grid 2,048	10% (6,000)	nothing	its fee, and its inventory, accrue RCAT
INVEST / WYFI	one-sided sell wall	10% (6,000)	nothing	likewise WYFI
INVEST / HIMS	one-sided sell wall	10% (6,000)	nothing	likewise HIMS
INVEST / DJT	one-sided sell wall	10% (6,000)	nothing	likewise DJT

The whole supply — 60,000 INVEST — is minted once into those five positions. The anchor's 36,000 against $10,000 sets an opening price of about $0.2778. Each wall offers its 6,000 INVEST from about $0.8333 — a $50,000 market cap, three times the open — upward to the top of the price range, and holds nothing on the other side, so below the wall there is simply no market.

Uniswap V4 charges a pool's fee on the input currency of a swap, so the fee is paid in both directions, in whatever the trader is spending:

• a buy of INVEST spends the pool's other token, so the fee accrues as USDG or one of the four stocks, depending on which pool the trade went through;
• a sell of INVEST spends INVEST, so the fee accrues as INVEST in all five.

That is half the point of the sell walls. RCAT, WYFI, HIMS and DJT are tokenized real-world assets the treasury wants to hold anyway. Normally it would have to buy them — and their order books are thin. Quoting a pool in them means every buy of INVEST pays the treasury in the asset itself, at no slippage at all. The other half is the inventory: the wall sells for that stock too, and that is by far the larger channel of the two.

A wall costs nothing to open. It takes no quote leg, so there is no stock for the protocol to buy before launch and no capital of its own at risk inside it — every dollar the owner put in went into the anchor.

Wallet-to-wallet transfers cost nothing, because nothing about the token levies anything. Nor does providing or withdrawing liquidity: a pool fee is charged on swaps.

The Treasury owns all five positions, and holds them differently on purpose:

• The anchor is one full-range position wrapped as a non-transferable share the Treasury holds 100% of. That share is registered as a reserve, valued conservatively off the USDG leg, and barred from redemption, so the protocol's own market cannot be redeemed out from under it.
• A wall has no share, no price feed and no reserve entry. It is registered with the Treasury as nothing at all. Its INVEST inventory is the protocol's own token, which is never counted as backing; what a wall earns becomes backing, what it holds does not.

Sweeping the accrued fees is permissionless — one call cranks all five, and every leg goes straight to the collector. A dormant wall cranks cleanly and returns nothing, which is the normal result at launch rather than a failure. Fees never rest in a vault and are never counted as backing while they sit there.

III b. A filled wall has to be banked

Collecting takes fees only. When a wall actually sells INVEST, the stock it was paid for sits inside the position as principal, where the fee sweep cannot see it. A second permissionless call takes it out, hands it to the Treasury by plain transfer — never as a deposit, which would mint INVEST against it — and re-mints the remaining inventory as a fresh one-sided wall at a higher price.

That re-setting is not bookkeeping. A partly-sold wall is a two-sided market: somebody could sell INVEST back into it and buy back the stock the protocol has just earned. Re-walling closes that in the same transaction, and the new wall can only ever move up.

It is also the point at which a wall's sale becomes backing, and nothing on-chain forces it to happen — it is a call somebody has to make, at least half an hour apart, off a price reference taken from the anchor rather than from the wall's own pool.

The fee is a property of the pool, not of the token

This is the one thing the design gives up, and it should be said plainly. A fee belongs to a venue. Anyone may open their own INVEST pool at a cheaper rate, or add liquidity alongside the protocol in this one and take a pro-rata slice of the 5%. And unlike a charge levied by the token itself, there is no lever that can reach out and pull a new venue into the fee — the fee lives in a pool key the protocol does not own for anybody else's pool, and no such lever could be built.

What bounds the leak is a market rather than a rule:

• At launch these are the only INVEST markets, and together they hold the entire supply. A competing pool cannot quote until somebody funds it, and funding it means buying INVEST through one of these first, at 5%.
• Depth beats a thin cheap pool. Routers split to the best all-in price, so a shallow pool at a lower fee loses to a deep pool at 5% for anything but dust — and the protocol-owned pools are the ones the protocol keeps putting backing behind.
• The anchor's tick spacing is deliberately the coarsest allowed. At 32,767 the narrowest position anybody can open in the anchor spans a price band of roughly 26×, which caps a third party's capital efficiency at about 2.5× full range. Nobody can mint a razor-thin position across spot and cheaply skim the flow the Treasury's own liquidity is earning.

The walls pay for that coarseness in a different currency, and the cost is stated rather than hidden. A wall cannot use the anchor's grid: one step of 32,767 ticks is a 26× price move, so a $50,000-market-cap wall would round either to several times its target or down to the launch price, where it would be emptied in the first blocks. The walls therefore use a grid of 2,048 — a 22.7% step — which does let a third party concentrate around the wall at roughly 20× full-range efficiency. Three things bound it: it is confined to the four wall pools and never touches the anchor; such a position is 22.7% wide while the protocol's runs to the top of the range; and building one means first buying INVEST from the anchor, which is exactly the price move the walls exist to sell into.

The walls do not split the anchor's depth — they hold no cash, so the anchor holds the entire seed. What they are is a claim on the supply rather than on the capital, and they are never used as a price reference and never sold into. They are an accumulation instrument, not a market.

What the protocol gets for that

• A token that can trade on Uniswap V4 at all. A token that takes a cut of its own transfers cannot settle inside the V4 singleton: the pool checks what it actually received, finds it short, and reverts the whole swap. Being a plain ERC-20 is the only way to be native to V4 and routable by every aggregator that indexes it.
• Buy-side revenue arrives in something the treasury wants. Buys in the anchor pay USDG, which needs no sale at all; buys in a wall pay the stock itself. That removes both the standing sell pressure of having to convert every unit of revenue and the backlog that accumulated whenever the oracle was out of band.
• No derating in the Uniswap app. A token detected as taking a cut of transfers is put behind exact-input-only swaps, a warning card, and a refusal to let users provide liquidity to it. None of that applies to a plain token whose venue happens to charge 5%.

Owning the liquidity — and growing it

Each position is seeded once, at launch, and the protocol never buys into it on the open market. The anchor can still be made deeper, and only ever out of reserves the Treasury already holds. Both paths are Boardroom decrees, and both end with the anchor's share back in the Treasury in the same transaction.

Neither path applies to a wall, and neither could: a wall has no share to issue pro-rata and no quote side to pair against. A wall grows only in the sense that it re-sets itself higher each time it is banked.

Moving a V2 position in. Uniswap V2 is still a live venue here: the lister creates and lists canonical INVEST/<reserve> V2 pairs as reserves in their own right, and the depository still opens bond markets quoted in those LP tokens — so LP bonds work exactly as they always did. What changed is where that LP ends up. An INVEST/USDG V2 LP sitting in the Treasury earns the V2 pool's fee for whoever trades there, not for the protocol. A decree can pull it out, burn it back into its two tokens, and add both to the V4 position at its live price. The V2 pool shrinks by exactly the Treasury's share of it; the protocol's own pool grows by the same value.

Deepening with USDG. The primary path, and the one to bond toward: Treasury USDG goes into the anchor's position paired with INVEST minted at the pool's live price. It mints supply, so it is a decree rather than an operator button, and it is bounded four ways at once:

• the rotator bucket — a rotation may draw at most a governed fraction of any one reserve's balance per rolling day;
• the excess-reserves gate — the mint is re-audited and refused past what backing already covers;
• a price floor stated in the proposal — the lowest INVEST price it is willing to mint at, quoted in the pool's own token rather than in dollars, so the opening figure is about 0.2778 USDG per INVEST and not 1.0. A decree sits in the Governor for days, so anyone who depresses the pool before it executes can only make the transaction revert, never make it mint more INVEST for the same deposit;
• a hard cap on the INVEST minted, which is the same bound in the unit a voter can check against supply.

Of those four, the excess-reserves gate is the one that binds at launch: excess is zero, so this path mints nothing at all until the gap described below has closed. It is armed and it will do nothing for as long as that is true.

Whatever the pool's ratio does not consume is settled the only two ways that leave nothing idle: leftover quote returns to the Treasury as plain backing, and leftover INVEST is burned — protocol-owned supply outside the pool is pure overhang.

The share itself can only ever move toward the Treasury, which is how the migrating contract hands its shares over inside the same transaction. It is also the reason no bond is offered against a pool share: a share is marked off its non-INVEST leg alone (the conservative mark that stops a bid-up INVEST price inflating the reserve valuation), so a two-sided contributor would be credited for roughly half of what they put in. The honest offer is the one above — bond USDG, or bond V2 LP, and let governance move it in.

III. Conversion — and the fees that are never converted

Of the six currencies the fees arrive in, only one is sold.

Fee currency	Where it came from	What happens to it
INVEST	sells in all five pools	sold for USDG — into the anchor pool only
USDG	buys in the anchor pool	already the currency the accumulator buys in: straight to the split, no trade
RCAT	buys in the RCAT wall	never sold. Straight to the Treasury, as RCAT
WYFI	buys in the WYFI wall	never sold. Straight to the Treasury, as WYFI
HIMS	buys in the HIMS wall	never sold. Straight to the Treasury, as HIMS
DJT	buys in the DJT wall	never sold. Straight to the Treasury, as DJT

Why the stock fees are kept rather than sold

Because selling them would be strictly worse in both directions at once. These are assets the accumulator would otherwise be buying with stablecoin, and buying them means paying spread in a thin book — measured within 2% of the price, the four range from a few thousand dollars of depth to a couple of hundred thousand. Being paid in the asset delivers the same thing at no slippage whatsoever, and the protocol never has to cross those books at all.

They arrive as plain backing: a transfer plus a re-audit, never a deposit, so no INVEST is minted against them.

The team's decaying share does apply to them — see stage IV.

Why INVEST is only ever sold into the anchor

A wall holds no cash. Below its wall price there is literally nothing on the other side of a sale, so a conversion routed at a wall is refused by name rather than filled badly. The anchor is the only pool that holds stablecoin, and it is the only place the protocol sells.

A sale through any other stablecoin pool — should one ever exist — additionally has to agree with the anchor's price within a governed band, or it is refused outright. That check compares dollars to dollars: each pool's median is translated through its own quote's price feed first, because a price quoted in RCAT is not a price in dollars.

Anyone can trigger a conversion — it is permissionless, and the protocol does not depend on a privileged party to run it. What a caller cannot do is choose the price.

Two bounds sit on every conversion, and a caller may only tighten them, never loosen them:

• A clip limit. A single conversion may sell at most 50 bps of the selling pool's INVEST leg, so one conversion moves that pool by roughly 50 bps at most.
• A price floor. The fill must land within 100 bps of a median price sampled over a window of thirty minutes to four hours. Spot price is never used.

The floor is computed net of the pool's own fee, read live at the moment of the call. A sale into a 5% pool fills about 5% below the middle by construction, so the fee is subtracted first and the 100 bps band is spent on price impact and noise rather than on the fee itself. In any case the fee is not a loss: it accrues to the Treasury's own position and comes back on the next collection.

The oracle behind the floor is a ring of sixteen observations of a protocol-owned pool's price, recorded at most once every thirty minutes; the answer is the median of those aged between thirty minutes and four hours. There is one such oracle per pool, and the anchor's is the one that anything outside this loop reads — the INVEST price shown in the app, the reference for bond pricing, and the reference a wall re-sets itself against, all come from the stablecoin pool and never from a pool quoted in an asset whose price the protocol itself publishes. It is fail-closed: with no valid observation in the window, conversion refuses rather than falling back to spot. Reviving it takes one permissionless observation and a thirty-minute wait.

A wall's own oracle is dormant, not broken. While a wall's pool is out of range there is nothing to sample and its median simply has no answer. That is the state every one of the four is in at launch, by construction, and it stays that way until INVEST is around three times the opening price.

A sampled median is not a cumulative time-weighted average, and it cannot be — a hookless Uniswap V4 pool keeps no running price for anyone to read. What bounds manipulation here is economics rather than averaging, and the bound is worth stating exactly:

• One poisoned sample never moves a median. Moving it takes a majority of the samples in the window, and each of those means a swap, a checkpoint and a swap back — a round trip that pays the pool's 5% fee twice, into the Treasury's own liquidity.
• The prize is capped. The only thing a depressed floor can be turned into is a single conversion of at most 50 bps of the position. The arithmetic works out at roughly ten times the cost of the benefit, before the attacker has paid for the majority of samples a median actually requires.
• Liveness, not theft, is the realistic failure. An attacker can keep the oracle out of band and stall conversions. That buys delay and nothing else, it costs the same fees continuously, and one honest observation plus thirty minutes undoes it.

There is no cooldown between conversions, because repeated conversions in the same block are self-limiting — each one pushes the pool further below the floor until the next one refuses.

IV. The split, and the team decay

Every USDG the collector holds — conversion proceeds and the buy-side fees that arrived as USDG already — is split between the team and the treasury on a fixed schedule that starts when the anchor pool is seeded and runs for thirty days. The stock fees are in this split too: the same decaying rate is applied in kind, so a buy in the RCAT wall pays the team its share in RCAT.

Days since launch	Team bps of the 500	Treasury bps
0	400	100
7.5	300	200
15	200	300
22.5	100	400
≥ 30	0	500

The decay is linear and continuous, not stepped; the table samples it. After thirty days the team share is zero forever and the entire fee funds the treasury. The clock is fixed at the anchor pool's seed and cannot be restarted by any later change to which pool is the anchor.

Stated plainly, because the table understates it. At launch that is 400 bps of the 500 — 80% of the fee flow going to the team rather than to backing, while backing is 85% short. It is why the volume needed to close the gap is five times larger before the vest decays than after. What makes the in-kind half of it cost nothing at the launch sizing is where the walls sit: they open three times above the launch price, so no wall earns a unit of stock until INVEST has already trebled — by which time the team's rate is zero.

Team compensation is this share and nothing else. There is no separate team token and no launch mint by default. The clock starts in the transaction that seeds the anchor pool, which is also the first moment anything can be traded — so there is no fee-free window and no head start. That transaction has not run: the clock has not started and nothing is tradable yet.

Nothing is reinvested. The collector routes the whole take and never adds proceeds back to any pool. Deepening a pool is a separate, explicit governance decision, never a side effect of a distribution; the flywheel's job is to turn fee flow into diversified backing rather than into more of the same pool.

V. Allocation — the accumulation desk

The treasury's share does not go straight to the accumulator. It lands at a small contract whose only job is to take the stablecoin that arrives and pass it on, split by weight, to destinations governance has approved. It exists so that what the protocol does with its revenue is a decision somebody makes explicitly, at a stated cadence, rather than a wire soldered at launch.

Who decides what

Who	Decides	Cannot
Governance — a vote, then the timelock	where money may go: each destination is approved one at a time; and the floor under reserve accumulation	—
The operator — one key, no delay	how much goes to each already-approved destination, any time	add a destination, move a token, or push reserve accumulation below the floor
Anyone — in practice a keeper, every seven minutes	when the split runs	anything else. Triggering a fixed-weight split is harmless, so it is permissionless and holds no seat

The floor is what makes the fast lane safe. It is 50% at launch: however the mix is re-weighted, at least half of the treasury's share goes to buying reserves, and only a vote changes that number. So the blast radius of a stolen operator key is bounded to shuffling money between places the protocol has already chosen — not to sending it anywhere new, and not to starving accumulation.

Two behaviours follow from the code and are worth stating because they surprise people:

• Revoking a destination does not stop it being paid. Approval and weight are separate facts; a revoked destination keeps its weight until someone re-sets the mix without it. De-escalating is therefore operator-first (instant) and governance-second (permanent).
• Raising the floor is not retroactive. It binds the next time the mix is set, so a floor change and a re-weighting belong in the same cycle.

At launch there is exactly one destination — reserve accumulation, at 100%. The desk is a pass-through until governance says otherwise, so nothing in section VI changes on day one.

What the audit sees

Every destination is registered with the Treasury as an address it trusts to be holding value on its behalf, so the audit prices what rests there through exactly the same fail-closed path it uses for reserves it custodies. Stablecoin waiting to be split is therefore backing, not a gap in the accounting. Two limits come with that, and both are deliberate:

• In-flight backing is not spendable backing. The Treasury can watch it; it cannot pay with it, and the redemption backstop pays only out of what the Treasury actually custodies.
• It is capped at 25% of all backing. Anything above the cap is simply not counted — never redistributed, and never a failure. That cap is the bound on how much of the protocol's backing can ever sit somewhere the Treasury only observes, and it is why each destination is a vote.

Any surface reporting backing has to show the split — custodied versus in flight — rather than one total. They are not interchangeable.

The second destination: a loss-less draw

The first destination beyond accumulation is a lottery over locked governance tokens where the deposit is never at risk. It is described in Treasury games; what matters here is that its prizes are the protocol spending part of its own revenue, and that turning it on took a vote to approve the destination plus an operator transaction to give it a weight. It can be switched off in the reverse order, the fast half first, without touching a single lock.

VI. Accumulation

The mandate

The basket is a position. It is weighted toward tokenized names the protocol wants the upside of, rather than spread for stability, and the machinery that makes that prudent is valuation, not selection: every reserve is discounted by a per-asset haircut before it counts as backing, and INVEST is minted only against the discounted excess. Because a haircut is a percentage, an appreciating reserve raises backing at the discounted rate while a fall was already assumed at listing — so the treasury can be long volatility without ever minting against a price spike.

Concentration is bounded by the machinery, not by convention: at most eight assets, weights summing to 100%, set by the operator and installable only for assets governance has already vetted, fed and routed.

The mechanism

The accumulation share arrives at the reserve accumulator as USDG — all of the treasury's share at launch, and never less than the floor set in stage V. The accumulator holds a weighted basket of up to eight reserve assets and buys that basket through Uniswap V4 routes that governance has pinned in advance. Each route is a specific, validated path — at most two hops, hookless unless a route explicitly opts in — so the accumulator cannot be steered into an arbitrary venue.

Every fill is floored against the Treasury's own price feeds, not against the pool it is trading in. A leg must deliver at least the feed value of what it spends, less a per-route tolerance band sized to cover the venue's own fees. The pool's price never sets the floor, which is what stops a manipulated venue from selling the treasury an overpriced asset.

Fills are additionally bounded by a per-fill size cap and a rolling daily cap per route, with tighter caps again when the asset being bought is priced by a manually posted feed rather than a Chainlink one.

A leg that cannot meet its floor is skipped, not forced. The stablecoin simply stays in the buffer for the next cycle. Nothing is lost, and anything left unspent for a day is swept into the Treasury as USDG rather than sitting idle.

Rotation

The same engine also runs rotation orders: an instruction to sell one existing reserve for another, either immediately, as a dollar-cost average, or spread over time. A rotation pulls the input out of the Treasury, swaps it under the same feed-based floor, and returns both the proceeds and any unspent input in the same transaction — so the Treasury is never left holding an IOU.

Rotations are bounded in three independent ways: by how much of any single reserve can be moved per day, by the same per-route caps accumulation uses, and by a daily ceiling on the marked value a rotation may cost the Treasury. That last one matters because reserves are marked after a haircut: rotating a 0%-haircut stablecoin into a 20%-haircut equity token books a marked loss even at a perfect fill, and the cap bounds how much of that the protocol will absorb in a day.

Because rotation is value-neutral inside its own transaction, it is deliberately governed by a different limit from the one that guards outright withdrawals — one that counts raw units of the token being rotated, so an inflated price on some unrelated reserve can never authorise rotating a real one away.

Why this funds rebases

Reserves grow. Supply does not. So the excess — treasury value above what already backs the circulating supply — grows with it, and staking emissions are capped by exactly that excess.

Trading volume becomes backing, and backing becomes emissions. That is the whole flywheel.

It also means emissions are honest about a quiet market: with no excess, a rebase mints nothing and does not fail. Staking never gets stuck, and the protocol never emits against reserves it does not have. At launch there is no excess at all — see below.

The opening gap, and the trade behind it

The protocol launches 85% unbacked, on purpose, and the launch transaction prints the figures rather than glossing them. (It has not run: today backing is $0 against a supply of 0.)

The protocol never counts its own INVEST as backing. The only thing it counts at t0 is the anchor's cash side, marked off the USDG leg with a 10% discount on top:

Total supply	60,000 INVEST
Opening price	~$0.2778 per INVEST — a ~$16,667 market cap
Sell-wall price	~$0.8333 — a $50,000 market cap, 3× the open
Counted backing	$9,000 — the anchor's $10,000 of USDG, discounted, and nothing else
Backing per INVEST	$0.15
The gap	$51,000 — 85%
Excess reserves	$0

Where the rest of the supply sits, exactly:

60,000 INVEST  =   9,000   counted backing   (the anchor's cash, discounted)
               +  26,000   the anchor's own INVEST side, which is not backing
               +   1,000   the 10% discount on that cash
               +  24,000   the four sell walls' inventory, which is not backing either

Three consequences, none of them hidden:

1. Every rebase pays exactly zero until backing exceeds the whole supply.
2. The genesis bond market is created and inert. Every market sits under a floor of $0.90 per INVEST — about 3.2× the opening price — so nothing fills until INVEST has risen that far. The floor is what stops bonds minting cheap INVEST into the gap; it is lowering-only and permanent, and lowering it would be the wrong response.
3. Deepening the anchor mints nothing, because it passes through the same excess gate.

What the shape buys in exchange

Nothing the treasury counts is marked at a price this protocol publishes. The one position it counts is quoted in a dollar stablecoin. A sell wall holds no purchased asset, so a badly posted stock price can only ever overstate an asset the protocol was paid — never one the owner funded. And a wall costs nothing to open: the entire seed went into the anchor.

The four stock prices are still numbers somebody posts, because those tokens have no independent price feed on this chain. What bounds a bad post is a budget the poster does not control — a limit per post, a limit per rolling day, a hard sanity band around the opening mark, and a 40–50% discount on the row on top of all of it.

How the gap closes

Two channels, and the fee lane is the smaller one:

• Fees. Closing $51,000 through fees alone would take roughly $5.1 million of volume at the launch team split, or $1.02 million once the team share has decayed to zero. That is an upper bound, because it counts fees only.
• The walls selling. A wall hands the treasury the stock it was sold for, not a fee on it. The first fourfold move above the wall delivers about $20,000 of stock across the four legs — twice the entire seed, and about two fifths of the gap. Half of a wall only sells at four times the wall price; ninety per cent needs a hundred times it.

Those figures describe what the walls would deliver if INVEST reached those prices. They are the shape of the inventory, not a forecast and not a projection of returns. At the launch price the walls have sold nothing.

What the mechanism will not do

• It will not mint INVEST. Fee proceeds are delivered as backing, never as a deposit, so the pool fee cannot inflate supply.
• It will not trade against spot. Every conversion is floored on a sampled median, and every basket fill on the Treasury's own feeds.
• It will not force a bad fill. A leg that cannot clear its floor is skipped and retried later.
• It will not let a hostile venue break the cycle. Each leg is isolated, so a misbehaving pool or hook can only make its own leg skip.
• It will not accumulate an asset governance has not vetted. Both ends of every route must already be enabled reserves with a live price feed.
• It will not let an operator redirect the revenue. The desk in stage V can only pay destinations governance approved one at a time, and never pays reserve accumulation less than the floor.
• It will not take a locked deposit. The one game the fee funds is loss-less by construction: it has no code path that sends a locked token anywhere but back to the address that locked it.
• It will not mark anything up because a posted stock price moved. No protocol-owned position is marked off a posted price at all: the one share the treasury counts is quoted in a stablecoin, and a sell wall is counted as nothing until what it earned has actually been handed over. A paused corporate action or a price outside its sane band takes that stock's own reserve row out of the accounting, and only that row.
• It will not treat a wall's inventory as backing. 24,000 of the 60,000 INVEST sit inside sell walls and appear in no treasury figure. Any surface reporting "the protocol-owned liquidity" has to carry that as a separate non-backing line rather than fold it in.
• It will not provide liquidity. The accumulator only swaps; it owns no position and has no path that adds to a pool. Neither does the collector: fee proceeds are routed, never put back into a pool, and the protocol-owned positions themselves are seeded once at genesis and then simply held. Liquidity the protocol comes to own later is a separate mechanism with its own bounds — LP bonds paid in discounted vesting INVEST, and a governed position vault that funds a bounded Uniswap V4 position from the Treasury's own reserves and marks it at the oracle-implied price. That vault is the one place in the protocol that adds liquidity after genesis, it can only ever mark backing down, and it re-proves that on every deposit; see Reserves and listing.

Two limits are worth knowing about. The first: stablecoin resting at the allocation desk, or at any destination it pays, counts as backing but is capped at 25% of the total and is not spendable by the Treasury — the split between custodied and in-flight backing is a number every surface has to show rather than fold together. The second: the accumulator and the position vault draw on the same daily allowance for each asset they pull out of the treasury. A large basket rotation and a large liquidity deposit in the same day can crowd each other out. Neither can take more than the treasury permits; they simply have to take turns.

Lineage

The conversion and split design is a port of NetNet, an existing protocol on Robinhood Chain whose contracts are verified on-chain. Investors Center carries over the same 500 bps rate, the same clip-and-floor-bounded conversion, the same decaying team split and the same conservative mark on protocol-owned liquidity, and adds the accumulator, the basket and the rotation engine on top.

What NetNet did not have is the anchor-and-walls shape: one market, quoted in a stablecoin, was the whole of it. Adding four one-sided pools quoted in the assets the treasury wants to accumulate — so that trading delivers them directly, and so that the inventory itself sells for them — is this protocol's own addition, and so is the discipline that comes with it: the stablecoin pool remains the price reference and the only place INVEST is ever sold, and the walls are registered with the Treasury as nothing at all.

NetNet's transfer tax was not carried over. NetNet charges 500 bps on every transfer that touches a keyed pair; Investors Center charges nothing on transfers at all and earns the same rate from its own pool instead. The reason is mechanical: a token that delivers less than it says cannot settle inside the Uniswap V4 singleton — the swap reverts outright — so a transfer tax and a V4-native protocol-owned market are mutually exclusive, and the market was judged worth more. The pair key, the exemption list and its four-day queue went with it; with nothing charged on transfers there is nothing to key and nothing to exempt. Several other NetNet mechanisms are also not ported: its liquidity valuation formula, its TWAP-premium emissions and its team option token. Emissions here stay capped by excess reserves.