How Rynex works

The trade

A market is an ordered pair. NVDA/SPY is how many SPY one NVDA buys. Long is a bet that NVDA beats SPY over your holding period and says nothing about whether either goes up.

Nothing is bought. You post USDG as margin, the contract records a notional and an entry mark, and at close it pays you the difference in USDG. The Uniswap v4 pools behind the two names are read, never traded.

Your profit is linear in the ratio. A long of N notional entered at R₀ holds N/R₀ units of the ratio, so it makes N · (R − R₀) / R₀, and the short makes the negative of that.

The holder gate

Opening a position requires holding 10,000 $RYNEX. The check is a plain balance read at the moment you open: the tokens are not staked, locked or taken, and they stay in your wallet the whole time.

Both the token and the threshold are immutable in the engine, fixed at deployment. The bar cannot be raised on anyone afterwards, which is the point: a gate the owner can tighten at will is a gate that can be closed behind you while you are inside.

What it does not gate

• Closing. Sell every token you hold and you can still get out. Nothing about the gate can trap you in a position.
• Adding margin. Making a position safer is never blocked.
• Depositing to the vault. Liquidity providers need none of it. They are the scarce side and gating them would only shrink the book.
• Liquidating. A liquidator who cannot act is worse than one who never held the token.

canOpen(address) answers it directly, and the desk asks the contract rather than working it out itself, so the page can never disagree with the thing that actually decides.

The mark

Each leg keeps an anchor: a price that anyone may step toward the pool's spot, by at most · basis points at a time and no more than once every ·. The mark of a pair is one anchor divided by the other.

The anchor exists because a spot read is not an oracle when funds depend on it. NVDA's pools hold roughly $6.5M of quote-side reserve, so a flash loan can shove the price 30% for the cost of the fees. Walking an anchor 30% instead takes hours of holding the pool away from its true price, paying the fee each time and handing everyone else a standing arbitrage in between.

The band goes on the ratio, not the legs

An anchor lags, so during a real move it is wrong in a knowable direction, and a market that halts whenever a leg runs ahead of its anchor halts on exactly the wrong day. When the whole tape drops 8%, both anchors are dragged at the same capped rate and the quotient barely moves.

So the deviation band is measured on the ratio, at · basis points, not on the legs. A pair is unstable when the two names separate faster than the anchors may follow, which is the only thing that can hurt a position here. Freshness is still checked per leg: cancellation is an argument about two anchors moving together and says nothing about one that stopped.

USDG cancels

Each leg is priced in USDG, so USDG appears once in the numerator and once in the denominator and divides out. An error common to both quotes, a depeg or a chain-wide mispricing of the unit itself, leaves the mark untouched.

Funding

A directional perp invents funding from two things: the cost of carrying the position and the crowd's lean. Here the carry is zero, and not approximately.

• Interest. Both legs are financed in the same USDG at the same rate. The differential is zero by inspection.
• Dividends. These are ERC-8056 tokens: a dividend is reinvested rather than paid out, so balanceOf never moves, uiMultiplier() steps up, and the pool reprices the token. Both marks are already total-return marks, so their ratio is a total-return ratio and there is nothing left to settle.

Rynex never reads uiMultiplier, and that is the point.

What remains is the lean. One rate, zero when longs and shorts are equal, rising with the imbalance to at most · a day at full one-sidedness, paid by the heavy side to the light side. Their notionals differ by exactly the skew, so the vault, which is what carries that skew, ends up paid the funding rate on precisely the risk it holds, with no second index and nothing credited before the money is there.

Leverage and the bell

Up to ·x while New York is trading. The underlying is shut for 136 of the week's 168 hours, and a pair holds up through that far better than a direction does because the two names gap together at the open and most of it cancels. Most is not all, and the residual is exactly what this book is exposed to, so the ceiling drops to 3x while the market is closed.

That is read from a calendar already deployed on chain, which works the New York date out of a raw timestamp, daylight saving and all, and holds every closure and early close through 2028. It applies to opening and to adding notional, never to holding. Nobody is force-closed at the bell.

Liquidation

Equity below · basis points of notional is liquidatable, and the liquidator takes half of that as the fee. The rest goes to the vault.

That threshold is read off the oracle rather than picked. Each leg's anchor moves at most 1% per five minutes, so the ratio moves at most about 2% in five minutes at the absolute worst, which makes a 5% buffer roughly twelve minutes of continuous adverse drift between a position becoming liquidatable and it becoming insolvent.

It is wide on purpose, because liquidation is barred while the pair is unstable. The cheapest liquidation to manufacture is one that should never have happened, and the buffer is what pays for closing that door.

Room is 1/leverage − maintenance, in the ratio, against you.

The exit

Closing is never barred. A protocol that can refuse to let you out has a worse problem than a stale mark, so close works while the pair is unstable, while a market is paused, and while the anchors are behind. Adding margin is never blocked either.

What it does instead is make instability unprofitable. When the pair is stable everyone settles at the mark. When it is not, the closer is marked at whichever of the mark and the spot ratio is worse for them: a long gets the lower, a short gets the higher. Someone whose position has genuinely gone bad can no longer duck out at a lagging anchor and hand the difference to the vault, and someone whose position is fine loses only the part of a favourable move the anchor has not reached yet, which they can wait out.

The vault

Deposit USDG, receive shares, and you are the counterparty to whatever the book does not net out on its own.

Longs and shorts offset each other first, and that matched notional costs the vault nothing. Only the residual skew is the vault's position, bounded by · against vault assets, so open interest can never exceed what the pot can cover. Withdrawals are refused if they would leave the vault holding less than the skew it is covering.

You earn the fees and the funding, and you pay the winners. It arrives in the share price, not the share count: your shares never change number.

The first deposit into an empty vault must be at least · USDG. Every deposit after that can be any size, because the share count can never return to zero. A thousand raw units of the opening deposit are burned to nobody, which is a tenth of a cent, charged once ever.

Fees

· basis points of notional on the way in and · on the way out. · goes to the treasury and the rest to the vault.

On a 20 USDG position that is 2 cents each way. There is no funding to pay if the book is balanced, and no fee at all for holding.

Waking a market

An anchor nobody steps for · goes stale, and the oracle stops quoting the pair rather than pricing off a number it no longer trusts. Anyone can fix it: one transaction steps every leg at once.

There is deliberately no background keeper. A step costs about 17 cents, so keeping every market fresh around the clock would run to four dollars a day at the bare minimum and forty-eight at the cadence the anchor design assumes. The desk shows a Wake it button when a market has gone to sleep, so the cost falls on whoever wants to trade and an idle market costs nobody anything.

A market outside the band is a different thing and shows no button: the two names are genuinely diverging faster than the anchors can follow, and the halt is the point.

Parameters

Read from the deployed contracts when you loaded this page.

Reading…

Contracts

Robinhood Chain, chain ID 4663. All unaudited.

The oracle's curator can still repoint a leg at a different pool. freeze() ends that permanently and has not been called, which is deliberate while the listings are still being checked, and is a live trust assumption until it is.

What can go wrong

• The contracts are unaudited. They have a test suite and a full lifecycle proven against a fork of this chain, which is not the same as an audit.
• The pools are thin. Open interest is capped at a quarter of the thinner leg's depth for that reason, but the anchor is what stands between a shoved pool and your position.
• A quiet market stops quoting. That fails safe, but it means you cannot open, and during a genuine divergence you cannot be liquidated either until the anchors catch up.
• Leverage does what leverage does. The liquidation mark is on the ticket before you open.
• The vault can lose. See above. Depositing is taking the other side of the book.

Nothing here is investment advice.