A floor that only ratchets up, and the price of building one

Genesis

The moment PAWL starts existing: the sale closes, the treasury is funded in USDG, the pool is created and trading opens. Every figure on this page described as “day one” is measured at that block.

Floor

The redemption value the contract will always pay, in USDG, at any size. It can rise. It cannot fall.

Ceiling

The price at or above which the protocol sells newly minted PAWL · the only way supply is ever created. It can rise. It cannot fall.

Band

The gap between the two, where ordinary trading happens. 400% at genesis, narrowing on a fixed schedule as reserves grow.

Backing

The USDG the protocol holds against the tokens outstanding. Committed backing is the 95% of it the floor is computed from; the rest is the solvency wedge.

Coverage

The floor as a percentage of the market price · how much of what you are holding is hard-redeemable. 85.5% at genesis.

The floor is denominated in USDG, not dollars. Every figure on this page is a quantity of USDG. If USDG loses its peg, the floor pays the same number of USDG and that is worth less. The protocol holds no other asset and makes no claim about the value of the one it holds.

·What it costs on day one

Published before the raise closes, and it does not depend on how large the raise is. Three separate things make it up, and two of them buy the holder something.

·The sale

Two bounds on the raise, two on a wallet. The floor does not move anywhere inside them · the marketing allocation is a fraction of tokens sold, so the raise cancels out and the maximum loss above is the same at every level. That is what makes it publishable before the window opens.

A single maximum wallet is 20% of the soft cap, so a small number of wallets can dominate the book. The maximum is also per wallet, not per person · thirty wallets defeat it. It is sybil-soft by construction: a courtesy, not a distribution guarantee.

The minimum excludes small contributions. A contributor with $100 cannot take part. “One price for everyone” now carries “above $500” beside it.

1The offer in one line

$0.10 per PAWL. All of it liquid at launch. No vesting, no lockup, no cliff.

No tiers, no bonding curve, no early-bird multiplier, no team round at a better price. One price for everybody, and you can redeem at the floor from the first block.

2The raise

The cap is deliberate. A small treasury is the phase in which a single purchase moves the floor most visibly.

How the sale is run, precisely. Contributions are collected manually to a named multisig whose address is published before the window opens, along with a running total. Distribution is handled by contracts deployed before the sale opens · so you can inspect the treasury that will hold the reserves in advance, and verify the funding transaction at close.

The pro-rata fill and the per-wallet caps are therefore commitments, not on-chain guarantees. No contract enforces them during collection. This is the only step in the system that requires trusting the team rather than checking code, and it is the step holding the funds.

The $10,000 maximum is per wallet, not per person · thirty wallets defeat it. It is sybil-soft by construction: a courtesy rather than a distribution guarantee. It also does not prevent concentration at the low end, since one maximum wallet is 20% of the soft cap.

The $500 minimum excludes small contributions. Collection is manual: every contribution is a human step to process, and below a few hundred dollars that costs more than it raises. A contributor with $100 cannot take part. “One price for everyone” now carries “above $500” beside it.

PAWL per ETH is not fixed, because the floor is denominated in USDG and ETH is not. Your ETH converts at deposit time, so nobody is exposed to ETH moving during the window.

Your entry price and your floor do not depend on how much we raise. The floor is $0.08550 either way, because the operating budget is funded outside the raise (see how the team is paid). That is deliberate: a budget taken out of the raise would make your floor depend on a number you don’t know when you commit.

PAWL launches on Robinhood Chain, where Uniswap v4 and its hooks are live.

3What you get

4Day one, at the $150k target

The floor is 85.5% of your entry price. Your maximum loss on day one is 14.50%, and it is published before you commit. It is a single figure with three components:

• The solvency wedge · 5.00 points. The protocol commits only 95% of what it holds. Not a fee, nobody receives it. It exists so the floor has margin, and it is the reason the floor can be committed to rather than estimated.
• Pool depth · 3.80 points. The cash seeding the pool sits outside the backing, so trading against the pool can never eat into the wedge.
• The marketing allocation · 5.70 points. Tokens minted at genesis as a fraction of tokens sold, and counted in the denominator the floor is divided by. It accounts for 39% of the day-one figure: without it the drawdown would be 8.80%.

For scale: in a typical token launch the equivalent figure is 100%, and it is rarely stated at all.

The 328% and the 80% are the same parameter written twice. A wide band means a long run before the protocol starts selling into demand · and it means a buyer at the ceiling is a long way above the floor. You never get one without the other, and we will never print one without the other.

The ceiling is a curve, not a wall. Each successive tranche the protocol sells costs more than the last, so sustained buying keeps moving the price instead of pinning it to one number. The ending price becomes the new ceiling, and it can never go back down.

5Where the numbers come from

Most of the figures on this page are derived rather than chosen, and the derivations are short enough to check on paper · a few of them on one line. They are here so the numbers are auditable · not so the mechanism is reproducible. What decides how fast the band closes, what each new tranche of supply costs and where the swap fee is routed is fixed at deploy and readable in the contracts; it is not published here.

The floor at genesis

f₀ = p × w × (1 − π) ÷ (1 + a)
   = $0.10 × 0.95 × 0.96 ÷ 1.06667
   = $0.08550

The allocation divides rather than subtracts because it adds tokens to the denominator without adding USDG to backing. And because all four terms are either a price or a fraction, the raise size cancels out · which is the whole reason a floor can be published before the window opens. Drop a and the same line gives $0.09120, the 8.80% drawdown quoted above; drop π as well and it gives $0.09500, the wedge alone.

The floor at any moment after that

f = B ÷ D
  = $136,800 ÷ 1,600,000
  = $0.08550

Committed backing over the floor denominator. Both are contract reads, so from launch onward you never have to take the floor on our word · B is USDG the treasury actually holds, and D is tokens sold plus the allocation. The coverage figure quoted anywhere on this page is just f divided by the price, and the band is the ceiling over the floor, less one.

Why somebody else leaving makes your floor go up

A redeemer hands back n tokens and is paid f for each. The denominator falls by n, but the reserve only falls by the committed part of what those tokens represented. What is left behind is the ratio the floor moves by:

f′ ÷ f = (D − w·n) ÷ (D − n)  >  1

Greater than one for any w below 1, at any size of n. That is the entire reason redemption has no queue, no capacity parameter and no spread: an exit is not a cost the protocol absorbs, it is the mechanism by which everyone who stayed gets richer. Redeeming 90% of supply in one transaction leaves backing-per-token higher than it started.

This is the structural difference from a bid wall. Olympus’ range-bound stability paid its floor out of a finite budget, so the floor was a bid that could be exhausted, and was. Here the floor is an accounting identity over a balance the protocol already holds: paying it reduces the liability faster than it reduces the backing. There is no budget to run down, which is why there is no capacity parameter and no regeneration timer to publish.

What the band width actually is

The band is not a presentational number. For a band of width c, the floor grows as a fixed power of backing, and the exponent is the band itself:

α = c ÷ (1 + c)
f ∝ B^α

at c = 400%:  α = 0.800  →  10× reserves gives 10^0.8 = 6.31× floor
at c = 200%:  α = 0.667  →  10× reserves gives 10^0.667 = 4.64× floor

One number wearing three hats. α is the floor exponent above; it is also the fraction of any ceiling fill that becomes accretion rather than paying for the tokens issued; and it is also the drawdown a buyer at that ceiling carries down to the floor. At the genesis band of 400% all three read 0.800 · which is where the 80% worst case quoted above comes from. That is why a wide band cannot be quoted as an upside figure on its own: it is the same arithmetic seen from the other side.

What a dollar of ceiling inflow buys

Along the ceiling path the relationship between money arriving and market cap collapses to a straight line · whatever the path, and whatever the tranche sizes it was delivered in:

MC = B₀ × (1 + c)  +  (k + 1) × m

m is cumulative ceiling inflow and k the slope of the ceiling at the time. Two readings, and both matter. The band width sets the starting line; the slope is the multiplier · every dollar through the ceiling adds k + 1 dollars of market cap, and nothing else in the design changes that rate. And the honest half of the same line: this is an exchange rate between inflow and market cap. It cannot manufacture the inflow, and no setting of any parameter will. Every milestone figure quoted on this page is this identity integrated, which is what makes them arithmetic rather than forecasts.

The most a single transaction can mint

Supply created in one transaction is capped at a fixed fraction of the denominator, immutable at deploy:

n_tx  ≤  ζ · D          ζ = 25%, immutable

binds when  m ÷ MC  =  ((1 + ζ)^(k+1) − 1) ÷ (k + 1)  =  28.125%

The binding point is closed-form rather than measured. The limit exists to bound the blast radius of a fault to a single transaction, and it does not bind on real orders: reaching it takes one buy worth about 29% of the entire market cap. Nobody mints their way through the schedule in one block, and that is arithmetic rather than a policy anyone is keeping to.

The smallest band the mint gate allows

The gate above has a consequence for the band that is easy to miss. A sale at the ceiling brings in w × (1 + c) of committed backing for every unit of floor claim it creates, so it can only clear the gate while

w · (1 + c)  >  1     ⇔     c  >  (1 − w) ÷ w  =  526.3 bps

Below a 5.263% band the protocol cannot issue at all · not because anyone decided that, but because the transaction fails its own gate. The scheduled minimum band clears that line by a wide margin, and it was checked against it rather than assumed. The same constant (1 − w) ÷ w is what the solvency wedge is worth as a fraction of committed backing, and it turns up a third time further down as the hard cap on the credit book · the wedge is a single 5% doing several jobs at once, and they compete.

The gate every mint has to pass

New PAWL exists only when someone buys at or above the ceiling. For m USDG arriving and n tokens issued, the transaction reverts unless

w·m  >  f·n

the sale must bring in more committed backing than the floor claim it creates. Minting is not a decision anyone takes and there is no key that overrides this; a mint that would lower the floor is not a policy we have chosen against, it is a transaction that fails. Every other guarantee on this page rests on this one line.

The end state, derived

Once the schedule finishes moving, the floor and the ceiling both scale as a power of the supply multiple, and the relationship between them settles at a constant. That is what makes the mature state arithmetic rather than a projection:

c∞   = (k + 1 − φ) ÷ ((1 − φ) × w) − 1  =  49.58%
cov∞ = ((1 − φ) × w) ÷ (k + 1 − φ)      =  66.85%

Two things worth noticing. The band and the coverage are one fact written twice · cov∞ = 1 ÷ (1 + c∞), so you cannot improve one without giving up the other. And neither depends on how large the protocol got: only on three constants fixed at deploy. That is the sense in which two thirds of the price being hard-redeemable is a guarantee about the shape of the end state, and says nothing whatever about whether the protocol reaches it.

What each new token does, forever

In the mature state the floor scales as uᵏ and market cap as uᵏ⁺¹ in the supply multiple u. So issuing another 10% of supply moves both by a fixed amount, at any size, permanently:

floor       1.10^0.40  =  +3.89%
market cap  1.10^1.40  =  +14.27%

Which is where the +3.9% and +14.3% quoted under where it can go come from. Growth never stops and never accelerates.

One clock, four dials

The band width, the swap fee, the staker share and the team share are not four independent schedules. They are four readings of one clock: a monotone high-water mark of committed reserves, running from $136,800 at genesis to $80,000,000 at maturity, in twelve steps. Every one of their launch and maturity values is already in the tables above.

Because the clock reads a high-water mark and nothing else, it has no reverse gear. Redemption shrinks the reserve but cannot move the protocol back to a wider band, a higher fee or a larger staker share, and there is no key that does it either. That is also why the fee cannot be gamed by pushing the price around: the only input is a number that has never been allowed to fall.

What is deliberately not here

Two things. The first is the shape of the interpolation · how each of the four dials travels between the launch value and the maturity value across those twelve steps, and what each successive tranche of supply costs on the way. The second is the calibration: why these endpoints and not others, which is the part that took the work and the part that never appears in the bytecode.

The schedule is not permanently private. It is fixed at deploy and readable on-chain from the first block, at which point every claim on this page can be checked against it. Withholding it before launch buys a launch window rather than a lasting advantage.

6About the 500,000,000 supply cap, and its $50M FDV

MAX_SUPPLY × $0.10 = $50,000,000 against a $150,000 raise. That ratio is 333×. Here is what sits behind it.

Start with the supply that exists. At genesis there are 1,660,000 PAWL · 1,500,000 sold to you, 100,000 as the marketing allocation, and 60,000 in the pool. That is 0.3320% of the cap. The other 99.6680% does not exist. There is no treasury holding it, no vesting contract, no multisig that can release it. It is not supply waiting to be dumped on you; it is not supply at all.

The one exception is the marketing allocation. It is minted at genesis, counted in the denominator the floor is divided by, and is the reason the day-one drawdown is 14.50% rather than 8.80%. It is not a claim on the reserve that appears later: it exists now, it is already in every figure on this page, and nothing can create another one.

Then the mechanism. The only code path that creates a PAWL is a ceiling fill: someone buying at or above the ceiling price, from the protocol, with the proceeds going into reserves. Every mint is gated on the sale bringing in more backing than the claim it creates, so a mint that would lower the floor reverts. Minting is not a decision anyone makes. It is a consequence of someone paying above the ceiling.

And then the bound most of this argument does not need. Well below the lifetime cap there is a second, tighter limit that applies at every block: circulating supply can never exceed genesis supply plus the backing the protocol has actually earned since, divided by a constant fixed at deploy. The constant is set from the weakest mint the protocol will ever accept, then halved, so a legitimate mint can never trip it. What it catches is the failure neither the gate nor the per-transaction limit does · a small over-mint, still accretive, compounding across thousands of fills.

Two things in that table are worth more than the headline. Headroom is not monotone and its minimum is not at genesis · walking the price from the floor to the ceiling entirely inside the band moves pool tokens into circulation while backing barely moves, which takes headroom from 1.09× down to 1.0675×. That is the tight point, it was found by testing rather than assumed, and a fuzz of mixed buy/sell/redeem paths found a global minimum of 1.0684× against it. The algebraic worst case, if every pool token entered circulation with no backing added at all, is 1.048×. No path breached the guard.

The honest consequence, because it cuts the other way too: the accurate claim is not “fixed maximum supply” but supply expands only against new backing and contracts when holders exit. That is a different sentence, arguably a stronger one, and integrators who filter on a fixed cap will notice the difference.

And the part that counts against the design:

• The cap is reachable in the project’s own terms. The modelled milestone schedule reaches it at roughly a $10.68B market cap, which is inside the range that model explores. It is not unreachable in practice.
• If it is reached, minting stops permanently and the ceiling is then defended only from tokens that redeemers have handed back. If that inventory empties, the ceiling stops being defended and price runs free above it. The floor is unaffected · it stays unconditional forever · but the upper bound becomes best-efforts.
• 333× is genuinely more dilution-shaped than 20×. Anyone whose instinct is “they are going to print 333 times what I bought” is reading it correctly. What they are missing is the price. Those tokens cannot be printed at $0.10. Reaching the cap requires $7.61 billion of cumulative buying at or above a ceiling that has risen the whole way, and at that point the floor is $14.23 against your $0.10 entry · 142× what you paid.

The number that means something is market cap against backing. On day one: $150,000 against $136,800. That ratio is checkable on-chain every day afterwards, it is what the protocol is actually promising, and it is the number we will publish. FDV measures overhang held by insiders; there are no insiders here and there is no overhang, so FDV measures nothing about this token except how large it could theoretically become.

If you want a reason to be suspicious of the cap, here is the real one, and it is not FDV. A fixed cap is a marketing and integration decision · screeners filter on it · and it is strictly worse for the mechanism than having no cap at all. It is survivable only because redeemed PAWL is parked and recycled rather than burned. That is a real trade, made knowingly.

7How the team is paid, and how stakers are paid

There is no team allocation. We contribute at $0.10 under the same $10,000 cap as everyone else. No free tokens, no discount, no separate round.

There is one allocation, and it is not ours: a marketing budget of 6.667% of tokens sold · 100,000 PAWL at the baseline raise, $10,000 at the entry price. It is minted at genesis and counted in the floor denominator, which is why the day-one drawdown is 14.50% rather than 8.80%. Of those 14.50 points, 5.70 are this allocation, 5.00 are the solvency wedge and 3.80 are pool depth.

We fund the build and launch costs ourselves, out of pocket, and recoup them from the protocol’s accretion fee · only if the protocol accretes. That is why your floor does not depend on how much we raise, and it means we are paid last and only if this works.

Ongoing, the protocol takes a share of ceiling accretion · the surplus created when the protocol sells above the floor. Never a fee on redemption, never a fee on the floor, never a fee on gross volume.

At launch, less than half of accretion reaches the floor. The staker share is front-loaded on purpose · the staked base is smallest at genesis, so a given share of accretion is loudest exactly when the protocol most needs holders to stay through the volatile phase. It decays on the same clock as the band, it is immutable, and no code path raises it. It is a cost to the floor in the early phase.

Three properties worth checking rather than trusting:

• Levied on accretion, not volume · it only exists when the protocol grows.
• Every rate is immutable or follows an immutable monotone schedule. There is no code path that raises either, including for us.
• A fee on gross inflow would become floor-destructive as the band narrows. This one cannot: it is accretive at any rate below 100%, by construction.

Separately, a scheduled in-band swap fee: 1.50% at launch, falling to 0.30% at maturity. On a buy the fee is taken from the USDG going in; on a sell it is taken from the USDG coming out. Both legs, same rate.

Where that fee goes, in full, because it is not what you would assume. 10% of it pays the team, always, from the first block and permanently. Of the remaining 90%: at launch it goes into the trading pool, until the pool holds cash equal to 10% of backing, and after that it goes to the treasury and does raise the floor. So at launch, none of the swap fee reaches the floor. Later, 90% of it does.

That is deliberate and it is a real cost. At launch the pool holds $6,000, which is thin enough that a $25,000 buy costs about 46% against mid and a few hundred dollars of selling walks the price all the way down to the floor. Routing the fee into the pool fixes that · a $25,000 buy costs about 12% instead · and it costs, measured over three years of simulated flow, about 19.6% of the backing the protocol would otherwise have accumulated.

The trade is a market that can absorb a trade at launch, against a floor that would otherwise be roughly 20% higher. It is fixed at deploy and cannot be changed afterwards.

And the team’s 10% is the one piece of our compensation that does not depend on the protocol working. Everything else we earn is a share of the surplus created when the protocol sells above the floor · zero in a flat market. This slice pays on trading volume instead, because the bills do not pause when the chart does. It costs the floor a measured 2.48% over three years at the central scenario. 10% is the ceiling we were willing to set, and it cannot be raised.

It is scheduled rather than flat because a flat 1.50% is defensible against a $6,000 pool and absurd against a $22M one · at scale it would be roughly 50× a competitive venue fee, and a mature PAWL charging it would not lose margin, it would lose the volume. The schedule reads only a monotone high-water mark of reserves, which nobody can push down.

8Staking

Stakers receive a share of ceiling accretion, paid in USDG · 30% at launch, decaying to zero at maturity. No token emissions, nothing to farm.

Read that last part again: it goes to zero, not to a small number. Staking is a launch-phase mechanism. It pays most when the protocol is young and volatile and you are most needed, and it stops entirely once the protocol matures into a savings instrument. If you are staking for the yield, the yield has an end date measured in reserves, not in time · it reaches zero when committed backing reaches $80M.

No APY is published · only the share, the pool, and a realised trailing figure computed from event data. A large share against a tiny staked base annualises to a number that is meaningless and that falls by an order of magnitude within months by construction. Any APY figure would be a forecast dressed as a rate.

It fades on purpose. The share decays on the same immutable schedule as the band, so as the protocol matures into a savings instrument the yield goes to zero on its own · by construction, not by us deciding to turn it down.

It stacks on the floor rather than replacing it. Your principal appreciates through the floor ratchet; the USDG is paid on top. Staking is not the reason to hold PAWL. It is what the protocol pays you for holding it.

No locks, ever. Stake and unstake whenever you want. Reward weight grows from 1.0× to 2.5× over nine months of continuous staking and resets if you withdraw · duration is rewarded, but nobody is trapped and there is no exit penalty. You can always redeem at the floor, staked or not. Staked tokens keep their full floor claim.

Staking is a transfer from non-stakers to stakers, and at launch it is a large one. The floor is lower for everyone and only stakers get the USDG back. That is what staking is in any design. Joining costs nothing · no lock, no minimum, no penalty and no permission · so the transfer is open to everyone holding the token.

9Borrowing against the floor

Pledge PAWL, receive 90% of its floor value in USDG, and keep every future ratchet on the tokens you pledged. There is no price oracle, no auction, no keeper and no market-price liquidation · credit is sized from the protocol’s own floor accounting, which is the same property that lets redemption work without an oracle. Because the floor cannot fall, a loan cannot go underwater from a price move.

The three numbers you should have before anyone pitches this to you, and two of them are limits rather than features.

The book is capped at 5.85% of supply, and not by choice. Lending out of the treasury reduces backing per token, and past a point redemption stops being accretive for everyone who stays. The condition is exactly one line · outstanding principal L against committed backing:

L  ≤  (1 − w) · A_liq        which at 90% loan-to-floor is 5.85% of supply

That is the solvency wedge again · the same (1 − w) that makes an exit accretive in the first place, now doing a fourth job and competing for the same 5%. Our own roadmap proposed capping it at a third of backing; measured against this line, a third is 6.3× too large, and the cap shipped is the wedge. The margin actually deployed is a fifth of the wedge, because other parts of the design already draw on the same allowance.

It adds at most about 0.32% a year to the floor. Interest paid in USDG raises backing with the denominator unchanged, so the ceiling on what the facility can contribute is the rate times (1 − w) · at the settled 8% rate and the shipped margin, 0.32% a year. It is real, it is guaranteed, and it costs a genesis buyer nothing. It is not yield and this page will not describe it as yield.

And it does not make the floor tick. Borrowing pushes the computed floor below the stored floor; interest paid refills that gap first, and only the surplus lifts the stored floor. On a book that never empties, cumulative interest has to exceed outstanding principal before the floor moves at all · 12.5 years at 8%, and measured between 10.4 and 12.7 years depending on how the book turns over. While a loan is open, the floor it was borrowed against is frozen, not falling. “A number that has risen every day since genesis” is not a claim this mechanism supports.

Three mechanics that follow from the accounting rather than from preference. Pledged tokens stay in circulation and stay in the floor denominator, so pledging cannot ratchet the floor on the way in and be extracted on the way out. A loan receivable is never counted as backing, and neither is accrued-unpaid interest · only USDG actually received. And surrendered collateral is parked in Inventory rather than burned, exactly like a redemption, which is what makes default trivially safe: the principal has already left backing, and the tokens leaving the denominator make the floor rise.

Positions are non-transferable · no debt token, no liquid receipt, nothing to rehypothecate outside the solvency model. Borrowing activates at the first maturity step rather than on a fresh threshold of its own, so it inherits a schedule that is already monotone, already immutable and already tested.

What this replaced is worth stating, because it is the more obvious idea. Putting reserves into a treasury-bill wrapper to make the floor compound fails in the direction it was meant to help: the haircut allowance is charged to the genesis floor whether or not a venue is ever used, taking the day-one drawdown from 14.50% to 18.77% · making the gap it was supposed to close larger before a cent is earned. It also takes redemption coverage from just above 1.0 to roughly 0.50, which would bring back the redemption queue this design removed. The facility gets a smaller version of the same objective with no venue, no haircut and no queue, and the smaller number is stated rather than dressed up.

10Where it can go

The floor rises when the protocol sells above it. Nothing else moves it up, and nothing moves it down.

No single figure is given, because the range is too wide for one to be meaningful. How much flow a token like this attracts is the input that moves outcomes more than every protocol parameter combined · measured, a factor of 3.4 on the headline floor from the flow assumption alone, against a widest-measured parameter effect of 30%. So here is the range, across a low, a central and a high flow scenario:

Each column is a distribution over simulated paths using real historical flow structure · three years of chain volume data, block-bootstrapped, so quiet stretches and drawdowns are real rather than invented. The columns differ only in how much flow arrives. These are model outputs, not forecasts, and nothing here is a promise about price.

The band we chose makes the early part better and the three-year number worse, and both halves are in this document. Against $500k of launch demand the price prints about 7.6× entry and the floor reaches 1.7×. Over three years of decaying flow the same width means more trading happens inside the band, where it only pays the swap fee, so the median floor is lower than a narrow band would have produced. That trade was made deliberately. If the launch is quiet, it will have been the wrong one.

Three things in that table deserve more attention than the headline:

• The band is still 236% wide at three years. The protocol has not matured into its savings phase inside the modelled horizon · not close. That is the direct cost of a band chosen for launch dynamics.
• Roughly 32% of the supply has been redeemed away. That is the mechanism working · every holder who left was paid the floor, and each exit raised backing per token for those who stayed · but it means the floor climbs partly because there are fewer holders, not only because reserves grew.
• The high scenario’s band is wider than the central one. More flow means the ceiling gets hit harder and the price runs further ahead of the floor. Growth does not close the band; the schedule does, and the schedule is keyed to reserves.

If demand never arrives, the floor stays where it is and remains redeemable against it. That is the downside case: a solvent protocol that does not grow, rather than a price going to zero.

Separately from any flow assumption, the coverage figure has a path that is arithmetic rather than modelled · it is what the ceiling integration says a given market cap costs, and it approaches the provable 66.85% from below and never returns:

Read it as an exchange rate, not a forecast. It states what each market cap costs in cumulative inflow and says nothing whatever about when, or whether, that inflow arrives · $694.9M is a large number and nothing here argues it is likely. Two honest notes on the table itself. Coverage is low early and that is a decision, not an accident: a wide launch band and thin early backing are the same choice seen twice. And the inflow column is partition-invariant while the floor column is not · the protocol fee is computed against the floor as of the start of each segment, so the floor and coverage figures are stated for realistic transaction sizes and are a lower bound only in that sense.

And the far end, for completeness. The protocol’s mature state is arithmetic rather than projection: a permanent 49.58% trading band with 66.85% of the price hard-redeemable, growing +3.9% of floor and +14.3% of market cap per 10% of supply issued, forever. That is a guarantee about the shape of the end state. It says nothing about whether the protocol gets there.

11What to weigh

The costs, limits and failure modes of the design, collected in one place.

• You are 14.50% above the floor on day one, and 5.70 of those points are the marketing allocation. A zero figure is not possible while a solvency wedge exists, but the wedge accounts for only 5.00 of the 14.50 points. Without the allocation the figure would be 8.80%.
• Someone buying at the ceiling carries an 80% worst case to the floor. That is the cost of a band this wide, and it is the same parameter that produces the 328% runway. It does not apply only to launch week: the schedule holds the band above 200% until roughly a $14M market cap, so a buyer arriving months later at the ceiling is still 70%+ above the floor.
• The protocol can stop growing, permanently, and still owe you the floor. Redemption raises the floor while the ceiling waits for its next maturity step, so a large enough exit can close the gap the mechanism needs in order to sell above the floor. One redemption leaving under 1.32% of the genesis denominator does it alone, and smaller ones compound toward it. Below that line no ceiling sale can complete and the protocol is redeem-only. Your floor is untouched: it is still fully backed and still pays at any size. What stops is the upside. Nothing prevents it, because preventing it would mean refusing a redemption.
• Coverage is low early, by design. At a $1M market cap about 22.3% of the price is hard-backed, and at $10M it is 29.2%. It crosses half somewhere between $100M and $1B. A wide launch band and thin early backing are one decision, not two.
• At launch only 45% of accretion reaches the floor, because of the staker and team shares. See how the team is paid.
• The floor is denominated in USDG, not in dollars. This is the single most important sentence on this page. Every PAWL redeems for a fixed and rising quantity of USDG. We do not price USDG against the dollar anywhere in the contracts · there is no oracle · so if USDG were to lose its peg, the floor would hold perfectly in USDG and you would lose value in dollars. The mechanism would be working exactly as designed while you lost money. Holding PAWL means holding USDG risk, transitively and in full. USDG is issued by Paxos Digital Singapore under MAS supervision, backed 1:1 by cash and short-term US government securities held at DBS Bank, and MiCA-compliant. Reserves are held in it and nothing else · no lending markets, no yield venues, no strategies. The 5% wedge is not depeg insurance; it covers valuation error and rounding. For reference, USDC · the most established stablecoin there is · traded at $0.87 for a period in March 2023. And the exposure grows: a $1B PAWL is a $559M USDG position.
• PAWL’s lifetime is bounded by USDG’s. The reserve asset cannot be changed by anyone, including us. If Paxos were ever wound down the way Binance’s BUSD was · peg intact, redemptions open, but the token retired by regulators over about twenty months · PAWL could not migrate to a replacement. Holders would redeem at the floor during that window and the protocol would end solvent.
• The pool is thin, and trades of a few thousand dollars move it a lot. At launch there is $6,000 of protocol-owned liquidity. A $1,000 buy costs roughly 18% against mid; a $5,000 buy costs roughly 85% and is mostly filled by the protocol at the ceiling rather than by the pool. Thin liquidity is what makes the floor the strongest number here, and it is also a cost to anyone trading in size. Whether it improves with scale is not guaranteed: the pool only deepens if trading happens inside the band, because it is funded out of the swap fee. If they do, it reaches 10% of backing · about $4.5M at a $100M market cap, and a $5,000 buy costs about 0.1%. If they do not, it stays at the $6,000 it was funded with at genesis, forever, and a $5,000 buy still costs 83% no matter how large the protocol gets. Both ends are published because the outcome depends on trading volume that cannot be guaranteed in advance.
• A wide band is worth more to a sandwich bot. Measured at these parameters, going from a 150% band to this one multiplies the best per-victim extraction by about 1.8×. Small in absolute dollars because the pool is small, but real, and a direct consequence of the band we chose.
• A large buy can briefly print above the ceiling. The protocol intercepts on the next transaction and the price comes back; the overshoot is bounded by whatever slippage the buyer accepted, not by the pool’s depth. It is a wick on the chart, not a break in the mechanism.
• Splitting a large buy into pieces gets you fewer tokens, not more · about 0.7% fewer on a $1M order, up to 2.4% on a $20M one, because the ceiling’s base price re-anchors at each reserve milestone. The difference stays in backing. It is a penalty on splitting, not an arbitrage against you. The direction was swept rather than argued: 7,200 comparisons across 100 fill sizes from $25k to $1B and 36 different partitions, and there is no case anywhere of the shape that would be an exploit · more tokens paired with a lower floor. The largest per-dollar advantage any partition obtains is exactly 0.000000%, and inside a single bucket the deviation is 0.000000000% on a $50k fill split 500 ways.
• While a loan against the floor is open, that floor is frozen, not falling. The credit facility contributes at most about 0.32% a year and, on a book that never empties, the first movement of the stored floor can be more than a decade out. See borrowing against the floor. It is capped at the solvency wedge for a reason, and the wedge is already doing three other jobs.
• Mature behaviour is a savings instrument, and the interesting phase is the early one.
• Nothing is live until genesis. The contracts are written and checked against the model, but none of this is running on-chain yet · so every figure here is computed rather than read from a deployment, and it stays that way until the pool opens.

12What cannot happen

Each of these is enforced in code, with no governance path around it, and checkable on-chain.

• The floor cannot decrease.
• The ceiling cannot decrease.
• Redemption at the floor works at any size · redeeming 90% of supply leaves backing-per-token higher than before, not lower. There is no bid wall to exhaust, no queue, no capacity parameter, and no redemption spread: you receive the full floor price.
• No PAWL is ever burned, and no PAWL is ever minted except by someone buying at or above the ceiling. Supply leaves circulation only by redemption, into a protocol account that can release it only back through the ceiling.
• Minting stops permanently at MAX_SUPPLY = 500,000,000, which is immutable. After that the floor is unchanged and unconditional; the ceiling is defended only from redeemed inventory, and becomes best-efforts if that inventory empties.
• A single transaction can never increase supply by more than 25%, and supply is bounded at every block by backing actually earned · two separate limits, both immutable.
• The treasury can never hold PAWL as backing, so backing can never be reflexive.
• No team allocation, no advisor allocation, no private round at a better price, no per-wallet cap exemption for anyone. There is exactly one token allocation and it is the marketing budget · 100,000 PAWL at the baseline raise, disclosed above with the 5.70 points of your drawdown it costs.
• No PAWL parameter can be changed after deployment. Not the fees, not the band, not the supply cap, not the reserve asset. Schedules are immutable and monotone: they can only move the way they were written to move.

The single exception, stated precisely: a multisig can pause ceiling issuance for the first 12 months. It cannot move funds, cannot mint, cannot lower the floor, and cannot pause the floor. That power is hard-coded to expire · after month 12 the protocol is unconditionally immutable, and the expiry block is checkable from day one.

That sentence is about PAWL’s own parameters and stops there. USDG, Uniswap, Robinhood Chain and its sequencer are external dependencies with their own governance, upgrade and pause powers.

One further precision, because the distinction is real: the floor is enforced in the canonical pool. Anyone can deploy a second PAWL pool without the hook, and price there can trade below the floor until arbitrage closes it · because the hook is the only mint path, that gap is a free trade for anyone who spots it. The accurate claim is “the floor holds in the canonical pool, and arbitrage restores it elsewhere,” not “PAWL can never trade below the floor anywhere.”

13You can leave on day one

A $1,000 contributor can redeem all 10,000 tokens at the floor immediately for $855 · 85.5% recoverable, in the first block, with no lockup and no spread.

The position is exitable at the floor from the moment it lists. What the design tests is not who contributes, but who stays once leaving is available at no penalty.