Design notes¶

In one paragraph

PUTGO has few rules, but each one is the result of a choice between several possibilities. This page records the reasoning behind those choices: why protection that can be bought at any moment is not eaten alive by buyers who only show up before bad news; why the premium is released to the pool second by second over the tenor rather than credited at purchase; why an early exit does not refund the premium; what auto-renew actually renews; which way each of the four rounding conventions points; what PUTGO explicitly is not; and which questions are still open. None of this changes any rule. The rules are on the product pages and on Markets and parameters.

Adverse selection: why buy-any-time holds up¶

Anyone who sells protection faces the same problem: the buyer knows better than the seller why they are buying now. The night before earnings, ahead of a macro release, or when a rumor about a protocol starts to spread, people who want protection arrive together. A seller who offers a fixed price at all times sells the most at precisely the moment it should sell the least. That is adverse selection.

PUTGO lets anyone buy at any moment, so it has to answer this head-on. It answers with five things stacked together, not with any one of them.

The rate rises with demand. The premium rate is not fixed; it is set by the Money Pool's utilization. When buyers pour in, reserved USDG grows, utilization rises and the rate climbs along the curve; past the kink at 90% it climbs more steeply; at 100% purchases stop. Someone trying to grab protection before bad news finds that the more they buy, the dearer it gets, and that there is always an end. The shape of the curve is in How the premium is priced.

The curve is recalibrated weekly from realized volatility. After a turbulent week, the next week's curve starts higher. This cannot foresee the future, but it keeps the premium in step with conditions rather than frozen at whatever number applied at launch. See Volatility, fair premium and weekly recalibration.

The floor is below the purchase price. The NVDA floor is 95% of the purchase price, the UNI floor 90%. The first 5% or 10% of any fall is the buyer's own. This works like a deductible: small moves do not trigger delivery, and nobody comes for a small payout that is almost certain to happen.

Only one official price at the settlement time counts. A momentary plunge during the day triggers nothing. A buyer cannot obtain delivery on the strength of "the price during one particular minute". See Why one settlement moment a day.

Every protection is fully reserved. Whatever the buyer's timing, the most the pool can pay out is the USDG already reserved. Adverse selection can make depositors earn less; it cannot make the pool owe money it does not have.

Where this design stops

Utilization pricing reacts to demand, not to news. When news breaks and the first buyers arrive, utilization has not yet had time to rise, and they pay the old rate. The curve's safety factor and the deductible are a buffer for that interval; they do not remove it. Depositors should treat this as a known source of risk, not a surprise.

Premium sniping and release over time¶

If the premium were credited to the pool in full at the moment of purchase, a simple arbitrage would appear: see a large protection about to be bought, deposit USDG just ahead of it, claim your share once the premium lands, then withdraw the USDG. That person carries almost none of the risk yet walks off with a slice of someone else's full week of compensation.

PUTGO's rule closes that route: the premium is released to the Money Pool linearly, second by second, from the moment of purchase to the moment of expiry. A depositor receives only the part released while they hold shares. Whoever has been in the pool longer receives more; someone who has just arrived receives only what has just been released.

Three further rules back this up:

Withdrawals wait. A withdrawal is requested first and executed only after the withdrawal delay (listed on Markets and parameters), and only from unreserved USDG. During the wait your shares still carry risk and still accrue premium.
Risk and reward walk in together. Newly deposited USDG carries, by share, the delivery risk of every live position from the instant it is deposited. You cannot take the premium without taking the risk.
The lump sum on early exit. When a buyer exits early, the part of the premium not yet released goes to the pool at once, to whoever holds shares at that moment. This is the one case in which premium is not released over time; it favors depositors, and the timing is the buyer's choice, which depositors cannot engineer.

The share price is accounted purely in USDG, and neither premium nor delivered assets enter the share value, so a premium landing in the pool does not push the share price up and impose an entry cost on later depositors. Together these rules make "drop in, take it, leave" unprofitable. The full depositor-side accounting is in Money Pool economics.

Early exit: why the premium is not refunded¶

The rule itself is short: you may exit at any time before expiry; the protected amount of the asset is returned immediately, the reserved USDG is freed immediately, the premium already paid is not refunded, the part not yet released goes to the pool at once, and there is no exit fee.

The premium is not refunded because the pool has been carrying the risk since the moment of purchase. Suppose the premium were refunded pro rata by days remaining: a buyer could buy before earnings, sit through the two most dangerous days, then exit and recover five days of premium. They would have paid only for the most dangerous window, while the pool carried the full delivery risk during it. That amounts to giving buyers the right to change their mind at any time for nothing, and that right is precisely the most valuable part.

Listed options handle this by letting the holder sell the option back at its market price, which includes the remaining time value. PUTGO has no secondary market and does not intend to assign a "current fair value" to every position. Between "no refund" and "refund an amount estimated by some model", PUTGO chose the former: the rule is simple, there is no valuation to dispute, and the tenor is only one week, so the premium itself is small.

The practical use is therefore clear:

If you need the asset, exit; it comes back immediately.
If you do not, letting the position run to expiry costs nothing more; on release the asset is returned anyway.
An exit is good for depositors, so there is no reason to charge an exit fee.

A void is not an exit

When a settlement batch is voided because the official price has been unavailable for too long or the chain has halted, the part of the premium not yet released is refunded to the buyer. The difference is that a void is not the buyer's choice; it is the protocol failing to deliver the settlement it promised. See How settlement works.

Auto-renew¶

Auto-renew is an optional switch on a position, set at purchase and changeable at any time while the position is live. When it is on and the position expires with a release, the returned asset immediately buys a new 7-day protection in the same market.

To understand it you need one sentence: every renewal is a new position with a new premium. Concretely:

The new purchase price is the official price at that settlement time, so the new floor moves with the price: if the asset rose, the floor rises; if it fell without breaching the old floor, the floor falls.
The new rate is read from the curve at that moment's utilization, and may be higher or lower than last week's.
The premium is taken again out of the returned asset, so the protected amount shrinks a little with each renewal.
USDG received through delivery is not renewed: you have already sold the asset at the floor, and there is nothing left to protect.
If the pool lacks capacity, the market is paused or the batch is voided, nothing is renewed and the asset becomes claimable.

One renewal (illustrative)

Last week's position: 9.92 NVDA, floor 209. Official price at expiry 231, release. Auto-renew is on, and suppose the rate at that moment's utilization is 1.2%.

New floor price      = 231 × 95% = 219.45 USDG
New premium          = 9.92 × 1.2% ≈ 0.119 NVDA
New protected amount ≈ 9.801 NVDA
New expiry           = the first 12:00 UTC after 7 full days from that settlement time

The floor moves up from 209 to 219.45, and you have paid about 0.119 NVDA for another week of protection.

The purpose of auto-renew is to make "protection every week" not require action every week. It changes no rule and gives a renewed position no discount. There is no limit on the number of renewals.

Rounding conventions, in plain words¶

Onchain amounts are integers in the smallest unit, and any division can leave a remainder that does not divide evenly. PUTGO fixes the direction of every rounding, on one principle: the remainder always goes to the party carrying the obligation, or always in the direction that keeps the pool able to pay. The four cases:

Amount	Direction	In plain words
Floor price	Down, to the price tick	The floor is allowed to be a hair lower; the pool's obligation never grows through rounding
Premium	Up	The last indivisible unit is paid by the buyer
Reserved USDG	Up	What the pool sets aside is always enough to pay a delivery
Void refunds, share amounts, withdrawal amounts	Down	What is paid back to an individual, and the shares minted, are never one unit too many

These remainders are single smallest units of a token and imperceptible in any transaction. The point of fixing the direction is not the amount but that nobody can farm the remainders by splitting transactions or repeating operations: every rounding points the same way, so the finer you split, the more it costs you.

What PUTGO is not¶

Saying plainly what a product is not often prevents more misunderstanding than saying what it is.

Not a perpetual. No leverage, no funding rate, no liquidation. A position lasts a fixed week and ends in exactly one of two outcomes, delivery or release. The most a buyer can lose is the premium plus the fall from the purchase price down to the floor (the first 5% for NVDA, 10% for UNI); the most depositors can pay out is the USDG already reserved.

Not a lending market. Depositors lend USDG to no one, and no one owes them interest. Their USDG may be converted into the asset at the floor on delivery; that is an asset-swap risk, not credit risk. The "rate" on the product page is the premium rate paid by protection buyers, not a deposit rate, and should not be read as an annual return.

Not an order book or an options exchange. No strikes to choose, no bids or asks, no partial fills, and there is no secondary market and no way to sell a position. Price comes from one public curve and capacity from the pool's USDG. The reasons are in Price protection as a product.

No institution stands behind it. The only source of funds for delivery is the USDG already reserved in the pool. The protocol provides no backstop capital, and the PUTGO team does not promise to make good any loss. That is exactly the point of full reservation: it needs no one to stand behind it.

Not a yield product. What depositors earn is the premium, paid in the protected asset. On delivery the pool also receives the asset in exchange for the USDG it pays out; that is a conversion of part of the deposit, not income. The USDG value of both moves with the market. No page will ever show a promised rate of return.

Not yet decided¶

A few product-level details are not yet final. They do not change the direction of any rule above, but until they are settled this site will not pretend they are.

The expiry boundary. Expiry is the first 12:00 UTC after 7 full days from purchase. If the purchase time plus 7 days lands on the exact second of 12:00 UTC, whether that day counts or expiry moves to the next day is not yet decided. The practical effect is that one second.
The official price source on mainnet. Which price source each market will use as its official price is not yet final. The price source used on the testnet is described in the Testnet guide.
The final scope of a pause. Currently a pause stops only new purchases and renewals; settlement, early exit, withdrawal and claiming continue, and that is how the testnet behaves. Whether this remains the final rule for deposits, withdrawals and settlement is still under review.
A per-wallet cap for depositors. Each pool has a total cap (see Markets and parameters); whether a separate cap per wallet will apply is not yet decided.
How new markets are approved. Only NVDA and UNI are planned at launch. Adding a market requires a separate review; the process has not been published and nothing is a commitment.

Once any of these is settled, Markets and parameters and the relevant product page are updated first, then this page.

Key takeaways¶

Buy-any-time holds up because five things stack: utilization pricing, weekly recalibration from volatility, the deductible of a floor below the purchase price, a single official price at the settlement time, and full reservation of every protection. Utilization pricing reacts to demand, not news; that is a known boundary.
The premium is released to the Money Pool second by second over the tenor; with the withdrawal delay and withdrawals limited to unreserved USDG, "drop in, take it, leave" does not pay.
Early exit is available at any time and returns the asset immediately, but the premium is not refunded: the pool has carried the risk since purchase, and a refund would hand out the right to change one's mind for nothing. A void is different, and there the unreleased premium is refunded.
Every auto-renewal is a new position with a new floor and a new premium; USDG received through delivery is not renewed.
Rounding directions are fixed: floor down, premium and reservation up, refunds and shares down. The amounts are imperceptible and the direction cannot be exploited.
PUTGO is not a perpetual, not a lending market and not an order book; no institution stands behind it; it is not a yield product.
The one-second expiry boundary, the mainnet price source, the final scope of a pause, a per-wallet cap and the approval process for new markets are not yet decided.