How it works, in pictures
The protocol page says what each contract does. This page shows the machine running: where the money from one trade goes, what happens to a launch in its first minutes, what a lot goes through, and what the treasury does in each kind of market. Who the people are, and what each one earns, is on who does what.
Follow the stock, then the profit
Start with one view about a stock: take profit in stages, keep the principal ready for a dip, and use the realized profit to buy back the strategy's token. Select a step below to follow the assets through that cycle.
This is an illustrative strategy, not a live fund or a performance record: +20% and +40% profit targets,
a 15% dip trigger, 20% of the current reserve per dip, no stop loss, and a 0.5% caller bounty.
STOCK and TOKEN are generic labels. These parameters and reward rates are not universal; read the selected
launch's own terms before acting.
A trade starts a new lot.
Someone sells
The pool settles the trade in stock.
The tax is shared
A portion goes to the treasury; other portions pay the creator, protocol and settlement caller.
A new lot begins
The received stock gets its own cost basis and profit targets.
Keep the two reward bases separate. At profit-taking, principal becomes USDG reserve. Of the profit retained in stock, 0.5% rewards the caller and 99.5% enters the buyback budget. A later buyback exchanges that stock for tokens; 0.5% of the tokens bought rewards the buyback caller and the remaining 99.5% is burned. Stock is spent to buy tokens, not burned itself. Actual amounts use contract rounding.
A dip uses a budget, not the whole reserve. Here the budget is 20% of the reserve at the time of the call; the dip caller's reward comes out of that budget before the rest buys stock. Each new lot has its own cost and profit targets. The dip reference starts from the first lot's cost and updates after a sale or a dip buy.
These steps need sufficient assets, valid prices, execution checks and someone to submit each transaction. With no stop, underwater lots can keep waiting and lose further value. Holding the token gives no treasury share or right to redeem its stock; a smaller token supply does not guarantee a higher price.
Rates are the factory's defaults as shipped (read on the deployed contracts, 2026-09-22): tax between 1% and 15% a side, 20% of the sell tax to the protocol, up to 30% to the creator, a 0.5% sweep tip and a 0.5% bounty, a sell spike of 90% decaying over 120 seconds, and a 99% buy window over 3 seconds. Every launch freezes its own copy at birth, so read that launch's live values before you trade. Prices in the examples are made up to make the arithmetic easy.
The whole machine
Four kinds of people, two contracts that hold money, and one place tokens go to die.
Read it as two loops. The trading loop (left): every buy burns a little of the token, every sell sends a little stock to the treasury. The treasury loop (right): a profit-taking sale returns principal as USDG to buy the next dip, and the gain funds token buybacks after rewards. An enabled stop can sell below cost. A keeper is whoever sends the next call first; nobody is appointed.
Follow one trade's money
A sell of $1,000, at a 10% tax, with the creator on 30%
The tax is taken in the stock, so a $1,000 sell pays $100 worth of the stock, and the seller receives the rest.
A buy of $1,000, at the same tax
For a standard exact-input buy, the buy tax is taken in the token and burned after the sweep tip. Exact-output buys use a different collection path; see the tax mechanism.
So the two sides do different jobs: buys shrink the supply, sells fill the treasury. A token that only gets bought never funds its treasury; a token that only gets sold funds a treasury whose burns are then the only buyers.
The first minutes of a launch
The creator's own first buy is the one thing a buyer should check before anything else: it is exempt from the window, uncapped, and visible on chain. See what to check. The diagram shows protocol-level options; the current website's Create form requires a developer purchase above zero. Treasury seeding is separate and remains optional.
The life of a lot
A lot is a parcel of stock with the price it was booked at. Each lot runs this on its own:
For a profit-taking sale, say a lot of 10 shares was booked at $100 per share and the price reaches $120. The rounded quantities below omit fees and slippage; stop-loss sales follow the separate rule below.
| shares | where it goes | |
|---|---|---|
| half the lot is sold | 5 | — |
| … of which the principal (what it cost) | 4.17 sold for $500 | the USDG reserve, for the next dip |
| … of which the profit | 0.83, kept as stock | after the take-profit bounty, funds token buybacks; bought tokens burn after the buyback bounty |
| the keeper's bounty | 0.5% of the profit | whoever sent takeProfit |
The retained stock profit is not turned into USDG. The token's pool is quoted in the stock, so the available profit buys tokens there. The bought tokens, less the buyback caller's reward, are burned.
One strategy, seven moves
A hypothetical price path using +20% / +40% take-profit targets, a 15% dip trigger, 20% dip allocation and no stop. These match the parameters associated with CRCLGRID, but this is not its trading history. Read its failed launch setup disclosure: a configured dip rule does not provide its own USDG reserve. The path below assumes booked stock, sufficient reserve for each dip, valid prices and callers.
The table follows which lot becomes eligible, rather than predicting balances or returns. Execution may require multiple calls; fees, caller rewards, price limits and actual fills affect the amounts.
| Step | Stock price | What the rule can do |
|---|---|---|
| 1 | $100 | Book the initial stock as L1, with a $100 cost and initial dip reference. |
| 2 | $120 | Take profit on L1's first half. Recovered principal funds reserve; stock profit after rewards funds buybacks. New stock tax is booked as L2 at $120. |
| 3 | $102 | A 15% fall from $120 permits a dip buy: create L3 and reset the reference to $102. |
| 4 | $86.70 | Another 15% fall permits a new dip buy, creating L4. L1's remaining half and L2 wait below cost because there is no stop. |
| 5 | $104.04 | L4 reaches its first target: $86.70 × 1.20. The cheaper lot can take profit before the older lots recover. |
| 6 | $122.40 | L3's first half and L4's remaining half are eligible for take-profit. Each lot uses its own cost. |
| 7 | $140 | L1's remaining half reaches its second target. L3's remaining half still waits for $142.80, and L2's first half for $144. |
A price path can create several separate opportunities, but no call is guaranteed to fill. Principal recovered from sales remains strategy capital. Stock spent on buybacks is exchanged for tokens; only the bought tokens, after the caller reward, are burned. A burn is not income paid to holders, and the token cannot be redeemed for the treasury's remaining assets.
What happens in each kind of market
These are possible behaviors of the no-stop example, assuming the relevant price, reserve, liquidity and execution conditions hold. Targets do not identify every market top or bottom. “No burns” below means no new buyback funding from the illustrated lots; existing profit may still fund buybacks, and buy-tax burns are separate.
| market | the treasury | burns | the creator | a keeper | a holder |
|---|---|---|---|---|---|
| Chops in a range | takes profit at eligible targets and buys eligible dips | when profit funds a buyback | paid on taxable sells | repeated calls may be eligible | repeated opportunities, without guaranteed returns |
| Grinds up | takes profit at each target; new tax books at ever-higher costs | steady | paid | steady work | burns, but the treasury holds less stock than if it had held |
| Falls and stays down | buys each 15% step with 20% of what is left; sells nothing | none | paid, if people still sell | only buyDip pays | no burn; lots wait for a recovery that may not come |
| Crash, then recovery | spends the reserve at the lows; the cheapest lots hit target first | a burst on the way back | paid | a burst of work | the dips pay out as burns |
| Flat | nothing triggers; new tax is booked, then waits | none | paid only from volume | nothing to send | nothing happens — holding earns nothing |
| Market closed (nights, weekends, holidays) | health() is false: no book, no sale, no dip | buy-backs continue, off the token pool's own price | paid — the token trades all week | buyback and sweep only | the token trades; the stock leg waits for the open |
| Feed stale, or a corporate action | the oracle fails closed; the stock leg pauses | buy-backs continue up to 5 days | paid | buyback only | as above |
| Heavy selling of the token | more tax arrives, more lots | the next buy-back spikes the sell rate | paid more | more to sweep and book | sellers pay the most right after a burn |
| No trading at all | nothing new arrives; what it holds still works its targets | only from existing profit | nothing | little | nothing is paid to anyone |
If it keeps falling
With this example's 20% allocation, each dip budgets a fixed share of what is left, so the reserve shrinks geometrically. This is not true of a strategy configured to spend 100%. With a 15% dip and a 20% lot:
| dip # | stock vs the reference before the first dip | this buy budgets | budgeted so far |
|---|---|---|---|
| 1 | −15% | 20% of the reserve | 20% |
| 2 | −28% | 16% | 36% |
| 3 | −39% | 12.8% | 49% |
| 4 | −48% | 10.2% | 59% |
| 5 | −56% | 8.2% | 67% |
| 10 | −80% | 2.7% | 89% |
The USDG budgets get smaller with repeated buys, and buying stops when execution would fall under the treasury's minimum lot. With no stop, underwater lots wait for their targets rather than trigger a stop-loss sale. They create no new buyback profit while waiting; previously available profit may still fund a buyback.
With a stop
A stop is the only way a lot sells below cost. It turns a waiting loss into a realised one: the proceeds return to the reserve as USDG, there is no profit, so nothing burns, and the keeper is paid from the proceeds. A stop never fires on a price only the pool vouches for — during a closure it waits for the open, because a fake weekend crash would otherwise lock in a loss that never happened.
Five numbers, four temperaments
The same stock, the same market, four different strategies. What each does is the table above, sharpened:
| Patient | Active | Range grid | Grid with a stop | |
|---|---|---|---|---|
| take profit | +100% / +250% | +8% / +18% | +20% / +40% | +20% / +40% |
| dip · lot | −12% · 5% | −5% · 20% | −15% · 20% | −15% · 20% |
| stop | none | none | none | −25% |
| chops in a range | rarely sells; almost no burns | sells and burns constantly | its home | its home |
| grinds up | sells late and big | sells out early, then only new tax | steady | steady |
| falls hard | waits, sips the reserve | spends the reserve quickly | waits, buys in steps | sells at −25%, burns nothing, rebuys lower |
| burns come | rarely, in large lumps | often, in small ones | once per swing | once per swing, none from a stop |
None of these is the right one. They are different bets on the same stock, published before anyone buys and frozen afterwards — which is exactly what lets a stranger choose between them.
What this picture does not show
- Holders have no claim on the treasury. No share, no NAV, no redemption. The only path from the treasury to a holder is the buy-back and burn.
- Burns are not a price floor. A smaller supply helps the price only if someone still wants the token.
- The keeper is not a guarantee. A call gets sent when its bounty is worth more than the gas; a tiny treasury may wait longer for its turn.
- The risks are listed in full on the protocol page.