In one minute
- $RELOAD is a token launched on Pons, a launchpad on Robinhood Chain. Every trade of it pays a 1% fee. The creator's share of that fee is the only money in this system.
- A contract splits that money three ways: 60% fills a pool of USDG (a dollar stablecoin), 30% buys $RELOAD and burns it, and 10% becomes a weekly prize.
- If you lock $RELOAD and link your perp trading accounts, and an exchange later force-liquidates one of them, you can claim part of the loss back from the pool, in USDG.
- The amount is limited three ways: by your loss, by what you locked, and by your share of that week's budget. The pool can never owe more than it holds.
- Attesters check each liquidation and look for hidden hedges before a claim can pass. They put up their own $RELOAD as a bond and lose it if they sign something false.
- People who lock the token vote on the settings, inside limits the code will not let anyone cross. Nobody can withdraw the pool: not the dev, not the voters.
The token
$RELOAD is created by Pons, not by us. It is Pons' standard token: it can be held, sent and burned, and that is all. It has no tax, no minting after launch, no blacklist and no pause. Everything else in this paper lives in separate contracts that hold and move tokens the same way any holder would.
How it launches
The whole supply is created in one transaction and placed on a Pons bonding curve, priced in USDG. A bonding curve works like a vending machine: it sells tokens at a price that rises as more are bought, and buys them back at a price that falls as they are sold. There is no presale and no team allocation. The dev holds no tokens.
| On the Pons curve (USDG launches) | Figure |
|---|---|
| Value of the whole supply at the first trade | about $3,236 |
| USDG paid in before the curve is sold out | $8,090 |
| Share of supply sold on the curve by then | about 71% |
| Value of the whole supply at that point | about $39,600 |
| Price at that point, compared with the first trade | about 12× |
Graduation
When the curve has sold its public share, anyone can trigger graduation. The $8,090 of USDG and about 20% of the supply become a Uniswap trading pool, opening at exactly the price the curve ended on. That liquidity is locked in a Pons contract for good. The remaining 8% or so of the supply is locked for good as well. From then on $RELOAD trades in that pool.
Our contracts follow this on their own. The price feed reads the curve before graduation and the pool after it, and the buyback buys from whichever one is live.
The trading fee
Pons charges 1% on every buy and every sell, on the curve and in the pool. Pons keeps 30% of that fee. The other 70% is the creator's share, and it is the money that funds everything described here. Pons also lets a creator add a second tax of up to 10% for themselves. $RELOAD launches with that set to zero.
Pons' own figures at the time of writing, read from its contracts. Pons fixes a token's fee terms on the day it launches. For the first seconds of any launch Pons also taxes buys heavily, falling to zero, to keep sniper bots from taking the opening.
Supply only goes down
Nothing here pays out $RELOAD. There is no inflation, no farming reward and no staking yield in the token. Three things remove tokens for good: the buyback, which burns what it buys; the bonds of attesters who are caught signing false claims; and the bonds of challengers whose challenges are rejected.
| Where tokens can be | What holds them there |
|---|---|
| In wallets | Nothing. Freely tradable. |
| Locked for rebates and votes | The staking contract, for 7, 30 or 90 days. The owner withdraws after the end date. |
| Bonded by attesters | The verifier contract. 45 days' notice to leave. |
| In the trading pool | Locked by Pons at graduation, permanently. |
| Burned | The dead address. Gone. |
Where the money comes from
Trading fees, and nothing else. No treasury, no token sales, no borrowed money. If nobody trades $RELOAD, nothing flows.
- A tradeSomeone buys or sells $RELOAD. Pons takes 1% of the trade in USDG.
- Pons' cutPons keeps 30% of that fee and credits the other 70% to the creator's balance in its fee escrow.
- CollectionA forwarder contract with no owner collects that balance and sends all of it to the Splitter. Anyone can press its button.
- The splitAnyone presses "distribute". The presser keeps 0.5% for the gas. The rest is divided as below.
- Reload pool 60%
- Buyback and burn 30%
- Survivor jackpot 10%
One hundred thousand dollars of trading, followed through
| Step | Amount |
|---|---|
| Trading volume | $100,000.00 |
| Pons' 1% fee | $1,000.00 |
| Kept by Pons (30%) | $300.00 |
| Creator's share, sent to the Splitter | $700.00 |
| Bounty to whoever pressed distribute (0.5%) | $3.50 |
| To the reload pool | $417.90 |
| To buyback and burn | $208.95 |
| To the survivor jackpot | $69.65 |
Limits on the split
Lockers can vote to change the split, but the pool always gets at least 50%, the burn and the jackpot always get at least 5% each, and the split can change at most once every 14 days.
If something other than USDG arrives
The Splitter can only pass on USDG. If it ever receives $RELOAD or another asset, a registered keeper sells it for USDG in small pieces. A sale cannot go through below a price floor: for $RELOAD the floor comes from the price feed, and for anything else lockers have to set one first. There is no "sweep" that sends the asset somewhere else.
Pons pays the creator's share to one address, the fee recipient, and only that address can collect it. Disclosure: the launch wallet holds that job first and keeps the first 1,000 USDG of creator fees to repay the launch costs (site, server, explorer API, gas); every transfer is visible on-chain, and the job then passes to the forwarder, one way. For $RELOAD the recipient is a forwarder contract with no owner. Anyone can press its button, the only place it can send money is the Splitter, and it has no way to change that or to hand the job to someone else.
Two things to know. The token is launched from an ordinary wallet, which is the recipient for the short time before our contracts are deployed; it then hands the role to the forwarder, and anyone can check on Pons that it has. And Pons' own contracts let Pons' owner reassign any token's fee recipient after three days' notice.
Locking
To take part you lock $RELOAD in the staking contract for 7, 30 or 90 days. Every lock is its own position, and three things are stamped on it for life: its dollar value, its rate, and its cap.
The dollar value, stamped once
The value is your tokens times the average price over the last 30 minutes. If that average is more than 20% away from the 24-hour average, the lock is refused, so nobody can lock at a price that was just pumped. After that the value never changes. If the price later halves or triples, the stamp stays the same.
Dollar size decides the cap
| Tier | Dollars at lock | Cap on a reload |
|---|---|---|
| Gold | $5,000 or more | 35% of the lock's value |
| Silver | $1,000 or more | 25% |
| Bronze | $250 or more | 15% |
| Under Bronze | under $250 | none: votes and jackpot only |
Length decides the rate and the votes
| Lock | Share of a loss it can cover | Vote and jackpot weight |
|---|---|---|
| 90 days | 35% | 2.5× |
| 30 days | 20% | 1.5× |
| 7 days | 10% | 1.0× |
Rules worth knowing
- Locks do not merge. Ten $999 locks are ten Bronze locks, not one Silver or Gold. A top-up is a new lock at the price of that moment.
- A wallet can hold up to 100 open locks.
- A lock under $250 is allowed. It votes and it counts for the jackpot. It adds nothing to a reload.
- After the end date you withdraw whenever you like. Nothing in the system can block or delay that: no vote, no pause, no penalty.
- A lock that has passed its end date keeps working, with the same coverage, votes and jackpot weight, until you withdraw it.
- If you withdraw while a reload is still paying out to you, you give up the part not yet paid.
- When lockers vote to change a rate or a cap, the change applies only to locks made afterwards.
Linking your trading accounts
A reload is only possible for an account you declared in advance. You list every perp account you control, and each link has to be proven.
- An account proves itself by signing a message with its own key. Nothing else is accepted: nobody, attesters included, can vouch for an account on your behalf, so nobody can link an account that is not theirs.
- That works today for exchanges whose accounts are Ethereum-style addresses. Exchanges on other kinds of chain are switched off until their accounts can sign for themselves in the same way.
- An account becomes claimable 3 days after it is linked. Your lock has to be that old too. This is the "lead time", and it is what stops someone buying coverage for a trade that is already going wrong. The one exception is the launch window: links and locks in place within six hours of the token's launch count from the moment that window closes, so the first people in are covered on day one. It happens once and cannot be reopened.
- A wallet can link up to 32 accounts. Unlinking takes 7 days, so an account cannot be dropped right before trouble.
Which exchanges count
| How hedges are found | Venues |
|---|---|
| Scanned automatically | Hyperliquid, Aster, Lighter, GMX v2, Ostium. Every wallet tied to yours is checked, whether you declared it or not. |
| Off at launch | dYdX v4, Drift, Jupiter Perps, Paradex. Their accounts are not Ethereum-style keys, and there is not yet a way for such an account to prove on-chain who controls it. Until there is, no account can be linked there. |
| Off at launch | Vertex, edgeX, Extended. Vertex has closed. The other two cannot be read without a credential that could also act on the account. The DAO can switch any venue on later. |
The rule behind the table: nothing is paid on what cannot be verified, and that covers both who owns the account and what happened on it. Aster shows positions but does not mark liquidations, so an Aster account counts for hedge checks and an Aster liquidation cannot be claimed.
A reload, start to finish
- Hour 0LiquidationAn exchange force-closes a position on an account you linked.
- Soon afterAttesters checkThey confirm it, scan for hedges and sign a claim. It appears in the app.
- Within 72 hoursYou submitOne transaction. The contract re-checks every rule and books the amount.
- End of the weekThe week settlesIf the week was over budget, every claim in it is scaled by the same factor.
- 72 hours from claimUSDG vestsPaid in a straight line. You collect as it unlocks.
Every one of these has to be true
- The account was on your declaration and linked at least 3 days before the liquidation, or by the end of the six-hour launch window.
- You have at least one lock of Bronze size or more, also at least 3 days old (or in place by the end of the launch window), that you have not withdrawn.
- The position was at least $500 in size and had been open at least 30 minutes.
- The exchange closed it. Closing it yourself does not count.
- The account really collapsed: its equity right after the liquidation is at most half of what it was before the position opened. Equity is taken as of the liquidation from the exchange's own records, so withdrawing afterwards changes nothing.
- You were not hedged for half or more of the position, and you did not open the opposite trade within an hour afterwards.
- The attesters scanned every venue the voters have marked as required.
- This liquidation has not been claimed before.
- You have had fewer than 2 reloads in the last 30 days.
- Neither your wallet nor that exchange account is banned.
- Enough attesters signed, with enough bonded behind them, and none of them is you.
If a claim is rejected you can send more information and ask again after 48 hours. The reason for every rejection is recorded.
How much you get
Three numbers are worked out, and you receive the smallest.
your loss, less any hedge × your rate
your locks' dollar value × their caps
your tokens ÷ all locked tokens × this week's budget
- A share of the loss. The loss is first reduced by how hedged you were. Then it is multiplied by the best rate among your qualifying locks. One small 90-day lock lifts the rate used for all of them.
- Your dollar cap. For each qualifying lock, its stamped dollar value times its own cap, added up.
- Your share of the week. Your qualifying tokens divided by every locked token, times the week's budget. It is counted in tokens, so splitting your stack across ten wallets gets you exactly the same.
The example: you lost $4,000 with no hedge. You hold one Silver lock stamped at $3,000 for 90 days, and your tokens are 40% of everything locked in a week with a $5,000 budget. The loss gives $4,000 × 35% = $1,400. The lock gives $3,000 × 25% = $750, Silver's cap. The week gives 40% × $5,000 = $2,000. You are booked $750.
The lifetime throttle
Once the reloads booked to your wallet add up to more than the dollars you have ever locked, your rate drops to the 7-day rate and your cap drops to the Bronze cap. Locking more value lifts it again.
Why it never pays to lose on purpose
The highest rate is 35% of a loss and the highest cap is 35% of what you locked. Whatever you do, you get back less than you lost and less than you put up. The rates and caps can be voted up, but never past 50% and 60%.
The weekly budget
The pool is rationed so that one bad week cannot empty it.
- A week is 7 days, counted from the moment the pool was created.
- At the start of each week the pool's balance is recorded. That week may pay out at most half of it. Fees that arrive during the week belong to the next one.
- Your claim is booked right away at the full formula, but nothing is paid until the week has closed.
- If the week's claims add up to more than the budget, every claim that week is cut by the same fraction until they fit. What is left over stays in the pool for later weeks.
- After the week closes, your USDG is released in a straight line over 72 hours, counted from when you claimed.
- A week that has closed is never changed.
The pool holds $20,000 when the week starts, so the budget is $10,000. Claims that week add up to $15,000. Everyone receives two thirds of what they were booked. The other $10,000 stays in the pool and next week starts from there, plus new fees.
The site shows a "coverage ratio": the pool's balance divided by all the dollars locked. It tells you how thin or thick the pool is. It is a description, not a promise.
Why it cannot be farmed
A pool that pays people for losing money attracts people who try to lose on purpose, or pretend to. These are the rules that make that a bad trade.
| Rule | What it stops |
|---|---|
| 3-day lead time | Buying coverage for a position that is already underwater. The six-hour launch window is the one exception, and it closes for good. |
| Equity must collapse | Staging a small liquidation inside a large, healthy account. |
| Rates and caps under 100% | Losing on purpose. You always get back less than you lost and less than you locked. |
| Share counted in tokens | Splitting one stack across many wallets to draw more from the week. |
| Averaged price at lock | Pumping the price for one block to stamp a bigger dollar value. |
| Lifetime throttle | Drawing more than you ever put up. |
| Hedge scan across venues | Being long on one exchange and short on another, and claiming for the side that lost. |
| Two reloads per 30 days | Repeating the trick quickly. |
| Size and age minimums | Dust positions and trades opened only to be liquidated at once. |
How hidden hedges are found
Attesters do not only look at the account you named. They look at the wallets your wallet paid in the last 90 days, the wallet that first funded yours, and that wallet's other recent payments. Only links your side created count: a stranger who sends you money is not treated as you, so nobody can make your claim look hedged, or get you flagged, by sending you dust. Sub-accounts on Hyperliquid and Lighter are read with the account they belong to, and so are coins held in spot there: a coin you hold against a short on it is a hedge. An opposite position anywhere in that group, while your position was open or within an hour either side, counts as a hedge.
Nothing is signed on a partial look. If an exchange that is switched on cannot be read for one of those wallets, or a wallet's history cannot be read in full, the claim waits and is checked again every half hour inside the 72-hour window. Things that keep an account unreadable are yours to fix: Aster's account privacy must be off, and an account with a very large trade history may not be readable at all. On GMX v2 and Ostium each position has its own margin: a position that was open at the liquidation and is changed before the claim is signed cannot be valued, and the claim is refused, so leave your other positions there as they are until it is signed.
Penalties
The looking continues for 30 days after a payout. If an account you never disclosed turns out to have been on the other side, your wallet is flagged. A flag takes back whatever has not yet been paid, records a strike, and bans the wallet and that exchange account from claiming for 90 days. A second strike is permanent. Unstaking always still works.
A flag can be contested. Lockers can vote to reverse it, which removes the strike and restores what was taken back.
The price feed
Three things need to know what $RELOAD is worth in dollars: a new lock, the buyback, and an attester's bond. They all read the same price feed, called the oracle.
How the price is recorded
About every ten minutes, an approved wallet called an observer presses "poke". The contract then reads the current price itself, from the Pons curve before graduation and from the trading pool after it, and writes it down. The observer chooses when to press. The observer never chooses the price.
From those readings the contract keeps two averages: one over the last 30 minutes and one over the last 24 hours.
| Used for | Which price | Why |
|---|---|---|
| Stamping a new lock | 30-minute average, refused if it is more than 20% from the 24-hour one | A short pump cannot buy a bigger tier. |
| The buyback's minimum | 30-minute average, less 3% | The buyback will not overpay into a manipulated price. |
| Selling $RELOAD fees for USDG | The highest of the two averages and the latest reading, less 3% | Fees are never dumped cheaply. |
| Valuing an attester's bond | The lower of the two averages | A pump does not make a bond count for more. |
When it stops
If nobody pokes for an hour, the price is treated as stale. New locks, buybacks and claim checks all wait until readings resume and half an hour of fresh ones has built up. Withdrawals are never affected. Any observer can also hit a pause if something looks wrong; only a vote can lift it.
Buyback and burn
30% of fees collect in the buyback contract as USDG. Its only use is to buy $RELOAD on the market and destroy it.
- Anyone can press "execute", at most once every 6 hours.
- Each press spends up to a set amount: $200 at launch, because the trading pool is shallow at first and a larger buy would move the price too far. Lockers can raise it as the pool deepens.
- The tokens bought go straight to the dead address. Nobody can retrieve them.
- The buy must receive at least the price feed's 30-minute average, less 3%. If the market cannot give that, the press fails and the money waits.
- Lockers can vote for a larger one-off buyback, but only using USDG already inside the contract.
- The USDG can leave in no other way. There is no withdraw.
The survivor jackpot
10% of fees build up as a prize pot. Once a period, a prize from it goes to one locker who was not reloaded during it. It rewards the people who lock and do not blow up. One draw pays at most the committers' bond, $10 at launch, and whatever the pot holds above that stays for the next draws. Lockers can vote the bond, and with it the prize, up as fees grow.
Who can win
- Any wallet with an active lock, including locks under $250.
- Not a wallet that received a reload since the period began.
- Your chances follow your weight: tokens times 1.0 for a 7-day lock, 1.5 for 30 days, 2.5 for 90 days.
- You have to be entered each period. Anyone can enter anyone, and a keeper bot enters every locker, so you do not need to do it yourself.
How the winner is picked fairly
This chain has no trustworthy random number, so the draw makes its own. Several people, called committers, each choose a secret number. During the period each one posts a locked fingerprint of their secret together with a USDG bond ($10 at launch). After the period ends they have 24 hours to reveal the secret and take the bond back. The winner is picked from all the revealed secrets mixed together.
Because every secret is locked in before any is shown, no committer can pick a number that produces a winner they like. One honest committer is enough to keep the result unpredictable.
One loophole is left in any draw of this kind. A committer who has seen everyone else's secret can refuse to reveal its own, which changes the result and costs it the bond. That is why a single draw never pays more than the bond: refusing can never be worth more than it costs.
- 7 daysThe periodLockers are entered. Committers post sealed secrets and bonds.
- Next 24 hoursRevealCommitters reveal and get their bonds back. An unrevealed bond is added to the pot.
- ThenDrawWith at least 3 reveals, anyone presses "draw". The winner is checked again.
- Any time afterCollectThe winner presses "claim prize" and receives the prize in USDG.
- If fewer than 3 committers reveal, there is no draw. The pot carries into the next period.
- Entries close when the period ends. If the wallet drawn has since withdrawn part of its lock or been reloaded, its weight is cut to what is still locked and the draw picks again. Nobody wins on tokens they have already taken back.
- The winner collects the prize rather than being sent it, so a payment that cannot go through never blocks a draw.
A known limit of this method: the last committer to reveal can see the result first and choose not to reveal, which changes it at the cost of their bond. The bond is what makes that expensive, and lockers can raise it, up to $100,000, as pots grow.
Attesters
A blockchain cannot see what happened on an exchange. Attesters are the people who look. They run software that watches linked accounts, confirms a liquidation, hunts for hedges, and signs a statement of what it found. A claim cannot pass without their signatures.
Anyone can be one
Nobody is appointed and nobody is voted in. You become an attester by bonding $RELOAD in the verifier contract and running the attester software. There is no minimum bond. A bond is a separate deposit from a lock: it earns no votes and no rebate coverage, and a lock can never be taken as a penalty.
A signature counts for what stands behind it
- Your bond is valued in dollars at the lower of the 30-minute and 24-hour average price.
- Half of that value counts. Lockers can set this anywhere from 20% to 100%.
- One attester counts for at most $50,000, however large the bond.
- A new bond starts at zero and grows to full weight over 7 days. Tokens added later start cold as well.
For a claim to pass, at least two attesters must sign, and what their bonds count for has to add up to at least what the pool would pay for the claim: the loss after the three caps, never more than the loss. So to push through a false claim that would pay $750, the signers must have at least $750 of counted bond at risk between them, and a successful challenge takes at least that much from them.
What happens to a false signature
- Anyone can challenge a claim that passed, by posting about $250 of $RELOAD and their evidence.
- Lockers vote on it. If the challenge holds, every attester who signed loses half their bond. The challenger gets their own bond back plus a fifth of what was taken. The rest is burned.
- If the challenge fails, the challenger's bond is burned. That is the price of a frivolous challenge.
- Lockers can also slash an attester directly when there is evidence.
- An attester who wants to leave waits 45 days for the bond, so penalties can still reach it. If a claim it signed is under challenge, the bond stays until the DAO has ruled.
Two more guards
- An attester's signature never counts on a claim for its own wallet.
- If the price feed is stale, no bond can be valued, so no claim passes until it is fresh again.
Attesters are not paid by the contracts. They bond because they hold the token and want claims to be honest. At launch the dev runs the first attesters; the software and the rules are the same for everyone who joins after.
Where this is headed
The launch set is small on purpose: a few known operators, each running a server and bonding its own tokens, while the software proves itself in public. The plan after launch is to grow the attester set the way proof-of-stake validator sets grow, into the shape Hyperliquid's has today: a few dozen operators and many thousands of people bonded behind them. That takes a second version of the verifier, and every step of it is a lockers' vote:
- Pay for signers. A slice of each week's budget to the attesters who signed that week's claims, split by the weight behind them, so running a node earns something.
- Delegation. Bond $RELOAD behind an operator from the app, with one signature and no server. Delegators share the operator's weight, its pay and its penalties.
- A minimum self-bond for operators, so an operator has its own tokens at risk before it takes anyone else's.
- Uptime rules. A node that keeps going silent stops counting, so a claim never waits on an operator who went away.
- Bonds in $RELOAD or USDG. An attester could bond either, or both. USDG would count at face value, since it cannot fall on a crash day; $RELOAD would keep counting at the conservative price with the haircut. Slashed USDG would go to the reload pool, slashed $RELOAD would burn as now. $RELOAD would stay the favoured bond in two ways that cost nothing in safety: it reaches full weight in 7 days where USDG would take 30, and weight backed by it would earn a larger share of signer pay.
Alongside it: a join page in the app for the on-chain part, and a one-line installer that turns a rented server into a node in minutes. The pool reads the verifier's address from the registry on every claim, so the DAO can vote the new verifier in without touching anything else, and until it does, the rules above stand unchanged.
Who does what
No company operates this. A handful of jobs keep it running, and most of them are open to anyone.
Lockers
Anyone holding $RELOAD- Does
- Locks tokens, links trading accounts, claims reloads, votes.
- Puts up
- Tokens locked for 7, 30 or 90 days.
- Gets
- Reload eligibility from $250 locked, votes after 7 days, a jackpot entry every period.
- Kept honest by
- The claim rules, hedge scans, strikes and bans.
Attesters
Anyone who bonds $RELOAD- Does
- Confirms liquidations, scans for hedges, signs claims, flags cheats found later.
- Puts up
- A $RELOAD bond, with 45 days' notice to withdraw.
- Gets
- Nothing from the contracts.
- Kept honest by
- Losing half the bond for a false signature. Cannot sign its own claims.
Challengers
Anyone- Does
- Disputes a claim that should not have passed, with evidence.
- Puts up
- About $250 of $RELOAD per challenge.
- Gets
- The bond back plus a fifth of the slashed bonds, if lockers uphold it.
- Kept honest by
- Losing the bond if the challenge is rejected.
Committers
Anyone with the bond in USDG ($10 at launch)- Does
- Supplies the randomness for the jackpot: posts a sealed secret, then reveals it.
- Puts up
- A USDG bond per period: $10 at launch, and the most one draw pays. Lockers can set it from $10 to $100,000.
- Gets
- The bond back on reveal. Nothing more.
- Kept honest by
- Secrets being sealed before any is shown. A bond not revealed in 24 hours goes to the pot.
Observers
Approved wallets. Lockers add or remove them.- Does
- Presses "poke" about every ten minutes so the price feed stays fresh.
- Puts up
- Gas.
- Gets
- Nothing.
- Kept honest by
- Never typing in a price: the contract reads it. An observer does choose the moment, so one with money could push the price just before recording it. A recorded price that strays more than 20% from the day's average stops new locks and buybacks. Other observers can pause the feed. Lockers can remove one.
Keepers
Anyone, usually a bot- Does
- Presses the public buttons: distribute the fees, run the buyback, enter lockers in the jackpot, draw it.
- Puts up
- Gas.
- Gets
- 0.5% of each fee distribution. The other buttons pay nothing.
- Kept honest by
- Each button doing exactly one thing, with the destination fixed in code.
The fee forwarder
A contract with no owner- Does
- Collects the creator's fee share from Pons and sends all of it to the Splitter, whenever anyone presses its button.
- Puts up
- Nothing. Whoever presses pays the gas.
- Gets
- Nothing. The dev takes no cut, apart from a disclosed first 1,000 USDG of creator fees that repays the launch costs and is collected by the launch wallet before the handover.
- Kept honest by
- Having one destination written into its code, with no way to change it or to pass the job on. The launch wallet holds the job only until it hands it over.
The Guardian
One multisig, for 180 days- Does
- Can cancel a vote that has passed but not yet taken effect. Nothing else.
- Puts up
- Nothing.
- Gets
- Nothing.
- Kept honest by
- Expiring for good 180 days after launch. It cannot propose, execute, or move funds.
At launch the dev runs the observers, the keeper bot, three committers and the first attesters, because nobody else will be there on day one. None of those jobs has special access beyond what is listed here.
Governance
The settings are controlled by the people who lock the token, and by nobody else.
Votes
- Only locked tokens vote. Unlocked tokens count for nothing, and the dev holds none.
- A lock starts voting when it is 7 days old. Its weight is tokens times 1.0, 1.5 or 2.5 for a 7, 30 or 90-day lock.
- Votes are counted as they stood when a proposal was created, so buying and locking after the fact changes nothing.
- Creating a proposal takes 0.5% of all voting power.
Two kinds of proposal
| Track | Needs | Used for |
|---|---|---|
| Standard | 8% of all votes taking part, and more than half in favour | Moving a setting inside its allowed range, venues, the fee split, pauses, attester penalties, reversing a flag. |
| Supermajority | 15% taking part, and at least 66.7% in favour | Moving a range itself, changing the voting rules, or replacing one of the supporting contracts. |
From vote to effect
- 1 dayNoticeThe proposal is public before voting opens.
- 5 daysVotingLockers vote for or against.
- 48 hoursWaiting roomA passed proposal sits in the timelock. The Guardian can cancel it during its first 180 days.
- ThenTakes effectAnyone can execute it.
Limits on the voters themselves
- A proposal can only call functions on a fixed list. Anything not on the list cannot even be put to a vote.
- Every setting has a hard range written into the code. A vote cannot go outside it.
- Each setting can change at most once every 14 days. Pauses are the exception.
- Changes to rates and caps apply only to new locks.
- A pause can stop new locks or new claims. It can never stop a withdrawal or a payout already under way.
What nobody can do
Not the dev, not the Guardian, not a unanimous vote. These are not promises. The functions do not exist.
- Withdraw, sweep or redirect the reload pool, the buyback money or the jackpot. Money leaves only as a reload, a burn or a prize.
- Mint, tax, blacklist or alter $RELOAD. The token is Pons', not ours.
- Stop or delay you withdrawing a lock after its end date.
- Change the dollar value, rate or cap on a lock that already exists.
- Rewrite a week that has closed, or a payout already booked.
- Slash a lock. Only attester bonds can be slashed.
- Pay a reload in $RELOAD. It is USDG or nothing.
- Replace the pool or the staking contract with a new version.
- Extend the Guardian past 180 days, or give it a second power.
- Vote with tokens that are not locked.
What can go wrong
Read this before locking anything.
- The pool may be small. It fills only from trading fees. With little trading there is little to pay out, and a reload can be scaled down to very little or nothing.
- The token can fall. Your tier is fixed in dollars when you lock, but what you get back at the end is tokens, at whatever they are worth then.
- Attesters read exchanges from outside. They cannot see inside a centralised exchange. A hedge there is invisible, which is why payouts are capped well below what was locked.
- Attester security follows the token price. Bonds are in $RELOAD. In a crash they count for less, exactly when liquidations pile up, so honest claims may have to wait for more bond or a recovery.
- The price feed needs someone to keep it alive. If the observers stop, new locks, buybacks and claims wait. Withdrawals do not.
- Observers are trusted with the price feed. A stolen observer wallet could nudge the recorded price by up to a fifth a day, which skews new locks and buybacks by that much, or pause the feed, which holds up claims until a vote lifts the pause. It cannot take funds or touch a withdrawal.
- Pons can redirect the fees. Pons' contracts let its owner change any token's fee recipient with three days' notice. And until the launch wallet hands the role to the forwarder contract, that wallet is the recipient.
- Not every venue counts. Five of the twelve exchanges are on at launch: Hyperliquid, Lighter, GMX v2, Ostium and Aster. On Aster a liquidation cannot be verified, so it is not paid. The others stay off until an account there can prove who owns it, or until it can be read safely.
- Nothing is paid on a partial look. If a switched-on exchange or a wallet's history cannot be read in full, or the account's equity at the liquidation cannot be established from the exchange's own records, the claim is not signed. It is tried again every half hour for 72 hours. Some causes do not clear: an account with an enormous trade history, or a GMX v2 or Ostium position that was changed after the liquidation.
- Early trading is thin. The pool opens with about $8,000 of depth. Prices move easily and buybacks are small until it grows.
- USDG has an issuer. Its issuer can freeze an address. That is outside anyone's control here.
- Robinhood runs the chain. Robinhood Chain is an Arbitrum Orbit chain with a single sequencer, operated by Robinhood, which decides which transactions get in and when. Arbitrum's forced-inclusion path lets a transaction the sequencer ignores be pushed through from the parent chain after a delay, so nothing can be locked away for good, but a claim's 72-hour window could pass in the meantime, and this protocol's transactions could be singled out.
- The code is unaudited. It is tested, and it has been rehearsed end to end, but no outside firm has reviewed it.
- Voters can make poor choices. Inside the hard ranges, a small turnout can still move a setting.
- A large majority can replace supporting contracts. Two-thirds of the vote, with 15% taking part, can swap the referee contract or the price feed for a new one. A dishonest replacement could approve false claims or waste the buyback fund. It takes a public vote and a two-day wait, and the Guardian can cancel it in the first 180 days. The pool and the staking contract can never be replaced. Early on, when little is locked, a large majority can be a small number of tokens, which is what the Guardian is for.
Every number in one place
"Fixed" means no vote can change it. Everything else moves only inside the range shown, at most once every 14 days.
Fees
| Setting | At launch | Range |
|---|---|---|
| Pons trading fee | 1% | set by Pons |
| Creator's share of that fee | 70% | set by Pons |
| To the reload pool | 60% | 50% or more |
| To buyback and burn | 30% | 5% or more |
| To the jackpot | 10% | 5% or more |
| Bounty for pressing distribute | 0.5% | fixed |
Locks and reloads
| Setting | At launch | Range |
|---|---|---|
| Lock lengths | 7, 30, 90 days | fixed |
| Rate by length | 10%, 20%, 35% | 5% to 40%, never above 50% |
| Tier thresholds | $250, $1,000, $5,000 | half to double |
| Cap by tier | 15%, 25%, 35% | 5% to 50%, never above 60% |
| Price check at lock | within 20% | 10% to 30% |
| Lead time for locks and links | 3 days | 3 to 14 days |
| Launch window (links and locks in place by its end count from its end) | 6 hours after the launch | fixed, once |
| Time to submit a claim | 72 hours | fixed |
| Minimum position size and age | $500, 30 minutes | fixed |
| Equity must fall to | half or less | 30% to 70% |
| Hedge that rejects a claim | half or more | 30% to 70% |
| Reloads per wallet | 2 per 30 days | fixed |
| Week length | 7 days | fixed |
| Share of the pool a week may pay | 50% | 25% to 50%, never above 60% |
| Payout period | 72 hours | fixed |
| Ban after a flag | 90 days; permanent at 2 | fixed |
| Open locks and linked accounts per wallet | 100 and 32 | fixed |
Buyback, jackpot and price feed
| Setting | At launch | Range |
|---|---|---|
| Time between buybacks | 6 hours | 6 hours to 7 days |
| Most spent per buyback | $200 | $100 to $10,000,000 |
| Allowance below the average price | 3% | 0.5% to 10% |
| Jackpot period | 7 days | 7 to 30 days |
| Committer bond | $10 | $10 to $100,000 |
| Reveals needed for a draw | 3 | 3 to 7 |
| Reveal window | 24 hours | fixed |
| Price readings | about every 10 minutes | at least 5 minutes apart |
| Price feed goes stale after | 1 hour | fixed |
Attesters
| Setting | At launch | Range |
|---|---|---|
| Minimum bond | none | fixed |
| Share of a bond's value that counts | 50% | 20% to 100% |
| Most one attester counts for | $50,000 | $5,000 to $1,000,000 |
| Signers needed | 2 | 2 to 9 |
| Bond needed against the payout | 100% | 50% to 300% |
| Time for a bond to reach full weight | 7 days | 7 to 90 days |
| Taken for a false signature | 50% of the bond | 10% to 100% |
| Challenger's reward | 20% of what is taken | 5% to 50% |
| Challenge bond | $250 of $RELOAD | $10 to $10,000 |
| Notice to withdraw a bond | 45 days | fixed |
Governance
| Setting | At launch | Range |
|---|---|---|
| Votes needed to propose | 0.5% | 0.1% to 2% |
| Notice before voting | 1 day | 1 hour to 7 days |
| Voting period | 5 days | 1 to 14 days |
| Turnout, standard | 8% | 4% to 20% |
| Turnout, supermajority | 15% | 10% to 30% |
| In favour, supermajority | 66.7% | 60% to 80% |
| Wait before a passed vote takes effect | 48 hours | 24 hours or more |
| Guardian's life | 180 days | fixed |
| Time before a lock can vote | 7 days | fixed |