Appearance
Settlement, Resolution & Void
A conditional market exists to answer one question: did the event happen? When that question is answered, every book in the family is disposed of in a single step — the conditional perpetuals settle or void, and the prediction binaries pay out. This doc explains exactly what each outcome means for each leg you hold, how to exit a conditional position before the event resolves, and how to position an open position through resolution.
If you are new to the instruments, read the conditional-markets overview first. The short version: every conditional market is a family of five linked order books built over one underlying perpetual and one binary event — the underlying perp, a conditional perpetual (YES) and conditional perpetual (NO), and a prediction binary (YES) and prediction binary (NO). The four event-driven books (the two conditional perps and the two prediction binaries) are the conditional legs.
Throughout, we use a generic example: an underlying asset trading around $100 with a scheduled event (say, an economic decision) that the family conditions on. Numbers are illustrative.
How an event resolves: YES, NO, or Void
An event has exactly three possible dispositions:
- YES — the event happened. The YES branch wins.
- NO — the event did not happen. The NO branch wins.
- Void — the event could not be resolved. Neither branch wins.
Resolution settles all five books in the family at once, in a single atomic step. There is no partial or sequential settlement: when the family resolves, every position in every leg is disposed of together, and the underlying perpetual keeps trading normally as the standalone perp it always was.
How the outcome is decided depends on how the market was set up:
- For some events, the outcome is derived automatically from an oracle — for example, a price-threshold event ("did the asset close above K?") reads the relevant oracle at resolution time and computes YES or NO directly. When an outcome is asserted for these markets, the exchange re-checks the assertion against the oracle and rejects it on a mismatch. You don't have to trust a hand-entered result; the machine recomputes it.
- For events without a clean machine-readable outcome, resolution is operator-asserted.
Void is special: it is never auto-derived. It is the escape hatch for an event that genuinely cannot be resolved — the data source failed, the event was cancelled, the question became ambiguous. Void always overrides any automatic path. You should treat Void as rare but always possible, because it changes the payoff of the conditional legs in a way that matters (see below).
Most well-constructed events make Void operationally improbable. A price-threshold event resolves to YES or NO as long as the oracle is alive; a scheduled decision resolves once the decision is made. But "improbable" is not "impossible," and the safe assumption when you hold conditional legs is that any of the three outcomes can land.
The lifecycle window
A family also has a deadline and a short resolution window after it. As an event approaches its deadline, the family transitions out of normal trading: the conditional legs freeze (no new orders) ahead of resolution, the outcome is settled if it can be determined, and if it still can't be determined after the resolution window the family auto-voids. Once a family is resolved, its conditional legs are closed for trading; only the underlying perpetual continues.
The practical takeaway: the conditional legs are not tradeable right up to the instant of resolution. There is a freeze first. Plan your exit before the freeze, not at it.
What YES/NO resolution pays each leg
When the event resolves YES or NO, the family pays out as follows. The winning branch is whichever matches the outcome (YES branch on a YES outcome, NO branch on a NO outcome).
The conditional perpetuals
- The winning conditional perpetual cash-settles at the underlying's mark. Your position is closed at the underlying perpetual's price at the moment of resolution, and your profit and loss is realized against your entry, exactly like closing a perp. If you are long the YES conditional perp at an entry of $101 and the event resolves YES with the underlying at $105, you realize roughly $4 per unit. The conditional perp's own trading price stops mattering at this point — settlement is to the underlying, because "the asset if the event is YES" simply becomes "the asset" once YES is known.
- The losing conditional perpetual is voided — see the next section. It does not settle to a loss.
This asymmetry is the whole point of a conditional perp. While the event is unresolved, the YES and NO conditional perps trade on separate books at separate prices (the conditional forwards). At resolution, exactly one of them converges to the real underlying and the other one disappears.
The prediction binaries
The prediction binaries are $0–$1 instruments whose price is the market-implied probability of their outcome. They pay out like classic binary options:
- The winning prediction binary pays $1 per contract.
- The losing prediction binary pays $0 per contract.
So on a YES outcome, the YES binary pays $1 and the NO binary pays $0; on a NO outcome, the reverse. A long binary that wins receives $1 against whatever it paid to enter; a long binary that loses keeps nothing. A short binary that wins (i.e. the branch it shorted lost) keeps the premium it received; a short binary that loses owes $1.
Worked example, YES outcome:
- You hold a long YES binary bought at $0.40. It pays $1. Profit ≈ $0.60 per contract.
- You hold a long NO binary bought at $0.58. It pays $0. Loss = the full $0.58 per contract.
Through any YES or NO resolution, the two prediction binaries together pay exactly $1 — one pays $1, the other $0. This is the binary box identity: the prices of the YES and NO binaries should sum to about $1 while trading (BY + BN ≈ $1), because at resolution exactly one dollar gets paid out across the pair. That identity holds through YES and NO. It does not hold through Void.
What Void means
Void is the outcome people get wrong, so state it precisely.
On a Void, the conditional perpetuals are voided and the prediction binaries pay $0.
Conditional perpetuals on Void: profit-and-loss is zero, margin is returned
A voided conditional perpetual is unwound with zero cash flow. Your profit and loss on it is 0 — not a loss, not a gain — and the margin you had posted against it is returned to your balance (net of any funding and fees accrued while the leg was open). It is as if the position never traded.
This is the single most important and most misunderstood property of conditional perps, so it is worth being blunt about what it is not:
- It is not a settle-to-$0. A voided conditional perp does not book a loss of (0 − entry) × size. There is no such loss. Whatever price you entered at, your realized P&L is zero.
- It is not a wipe. You do not lose your position's value; you get your margin back and walk away flat on that leg.
The same "voided, margin returned" treatment applies to the losing conditional perp on a normal YES/NO resolution. When YES wins, the NO conditional perp doesn't settle to a loss — it voids, P&L zero, margin returned. When the event Voids outright, both conditional perps void this way.
So a conditional perp has only two possible fates: either its branch wins (and it cash-settles to the underlying like a perp) or it voids (its branch loses, or the event Voids — P&L zero, margin returned). It never settles to a loss for losing its branch; losing its branch means it never happened.
Prediction binaries on Void: both pay $0
The prediction binaries are different. On a Void, both the YES binary and the NO binary pay $0. The exchange honors neither side's bet.
This is the case where the binary box breaks. Through YES or NO, the YES and NO binaries together pay $1. Through a Void, they together pay $0. If you bought a binary outright and the event Voids, you lose your full premium — there is no refund of the binary's price the way there is a margin return on a conditional perp. (Binaries minted at a $0 entry through the cash-out path, described below, have zero P&L on a void regardless, because they were entered at $0.)
Putting the three outcomes side by side, per contract:
- YES outcome: YES binary → $1, NO binary → $0. YES conditional perp → settles at underlying; NO conditional perp → voids, margin returned.
- NO outcome: NO binary → $1, YES binary → $0. NO conditional perp → settles at underlying; YES conditional perp → voids, margin returned.
- Void: both binaries → $0. Both conditional perps → void, margin returned.
The asymmetry between the two instrument types on a Void is deliberate. A conditional perp is a margined, two-sided position — voiding it with margin returned is the neutral, fair unwind. A prediction binary is a pre-paid directional bet on the event happening — if the event is declared unresolvable, the bet is refused on both sides and the premium is forfeited. Price binaries on families where a Void is plausible with that in mind: a sum-of-binaries that looks like a small free edge is not free if the family can Void to $0 on both legs.
Closing a conditional perpetual early: the cash-out mechanic
This document is the canonical reference for how early exit from a conditional perpetual works. You don't have to wait for resolution to exit a conditional position, but exiting a conditional perp early works differently from closing a plain perp, and the difference is structural, not a quirk. The whole mechanic follows from one fact: cash only exists once the event settles. Until then, your conditional profit and loss is a claim on the event, and the tradeable form of that claim is a prediction binary — so that is exactly what an early close hands you.
Why conditional P&L isn't paid as cash
A conditional perp's gain is a branch-contingent claim. "I'm up $5 per unit on the YES conditional perp" really means "I have a $5 claim if the event resolves YES, and nothing if it resolves NO." That value lives entirely inside the YES universe. There is no pool of unconditional cash standing behind it, because the cash that backs the claim is only created when the event settles and exactly one branch is paid out. Hand it to you as cash today and the books wouldn't balance if the other branch wins — the claim you were paid for would simply cease to exist.
So closing a conditional perp before its event resolves does not pay cash. Instead it crystallizes your realized conditional P&L into the instrument that is the tradeable claim on that same event: a prediction-binary position, entered at $0.
What an early close issues you
When you close a conditional perp before resolution, the exchange issues a prediction-binary position at a $0 entry, in the firing branch of the conditional perp you closed — "firing branch" meaning the branch under which that conditional perp pays (the YES binary for a YES conditional perp, the NO binary for a NO conditional perp). The direction follows the sign of your realized P&L:
- Close at a gain → a long position in that branch's prediction binary, at a $0 entry.
- Close at a loss → a short position in that branch's prediction binary, at a $0 entry.
The size is the magnitude of your realized conditional P&L expressed in binary tokens — roughly one token per $1 of P&L (illustrative granularity). Because the entry is $0, each token carries the full payoff of its branch: a long $0-entry binary that wins pays the full $1; a short $0-entry binary that wins keeps the full $1. The position is your conditional claim, now written as a binary on the same event.
Very small amounts round to zero. Below roughly a cent per token, or under about a dollar of conditional P&L, there is nothing to issue and the close simply realizes zero — there is no fractional-token dust. Close in meaningful size for the cash-out to carry value.
How that position turns into money
Once you hold the $0-entry binary, you have two clean paths to value, plus one shortcut:
Hold it to resolution. The binary resolves with the event — $1 if its branch wins, $0 if it loses (and $0 on a Void; see What Void means above). A long that wins collects the full $1 per token; a short that wins keeps the full $1. This is the patient path: you carry the crystallized claim through the event and it pays out at settlement exactly like any other prediction binary.
Sell it on the prediction-binary book today. The binary trades at roughly the implied probability of its branch, so selling realizes about (implied probability) × tokens in cash right now. Close a profitable YES conditional perp into 100 long YES tokens, and if the YES binary is quoted around $0.62, selling them returns roughly $62 in cash today (illustrative). This is the impatient path, and it is sourced from the binary book — immediate early cash-out depends on liquidity in the prediction-binary leg, because that book is where the claim converts to cash ahead of the event.
The shortcut: if you already hold the opposing binary, closing realizes cash directly. When the $0-entry binary the close issues nets against a binary you already hold, the two collapse into a settled cash result on the spot — you don't route back through the book, because the offsetting binary you're holding is the other side of the claim. A long firing-branch binary issued into an existing short of that same binary (or the reverse) cancels, and the realized value is paid out as cash immediately.
Why this is one mechanic, not a trick
All three paths are the same instrument viewed at different moments. The early close fixes the magnitude of your conditional claim and re-expresses it as a $0-entry binary; from there the binary either resolves with the event, sells on its book at the implied probability, or — if you're already holding its opposite — settles to cash immediately. There is no separate "mint cash from a conditional" primitive and no split/merge step. The binary is the cash-out, because the binary is the tradeable claim on your conditional P&L, and cash is simply what that claim becomes once it meets the event or its opposite.
Why you close first, not hedge-then-close
If you tried to lock in a conditional gain by selling binaries while leaving the conditional perp open, the lock would not hold: the open conditional perp keeps marking and moving, while the binary you sold is fixed. Closing the conditional perp first freezes the magnitude you're converting into a $0-entry binary, so any binary trade exactly offsets it. Close, then convert — never the reverse.
The lifecycle, from open to settled
Putting it together, the life of a conditional perp position:
Open. You buy (or sell) the conditional perp on its own book at the conditional forward — the price for "the asset if the event resolves this way," which generally differs from the plain underlying price. You post initial margin; the position is margined through the exchange's worst-case scenario engine, which requires you to be solvent under every resolution outcome.
Hold. The position marks on its own book and shares your single cross-margin balance with everything else you hold. Conditional perps and prediction binaries carry no funding — funding applies to perpetuals only. Your conditional P&L is a branch-contingent claim, not withdrawable cash, until you either cash out or the event resolves.
Exit early (optional). Close the conditional perp before resolution. This issues a $0-entry prediction-binary position in the firing branch (long on a gain, short on a loss), as described in Closing a conditional perpetual early: the cash-out mechanic above. From there: hold it to resolution, sell it on the binary book at the implied probability for cash today, or — if you already hold the opposing binary — realize cash on the spot.
Pre-resolution freeze. As the deadline approaches, the conditional legs freeze ahead of resolution. New orders on the conditional legs stop here.
Resolve. The event resolves YES, NO, or Void, and all five books settle atomically. Your winning conditional perp cash-settles at the underlying mark; a losing or voided conditional perp returns margin at zero P&L; binaries pay $1 / $0 / $0 as above.
Settled. The conditional legs are closed. The underlying perpetual keeps trading.
Flatten before resolution: positioning across the branches
This is the canonical statement of how to carry an open conditional position into resolution, and the most important positioning fact in this document — counter-intuitive, and concerning the position people think is safest: the "hedged" one.
How a conditional-perp-plus-perp hedge behaves across branches
A common construction is a conditional perp hedged with an offsetting underlying-perp position — for example, long the YES conditional perp and short the underlying perp, expecting the two legs to net out. That hedge holds in the branch where the conditional perp wins.
Walk through the Void branch (or the losing branch). Suppose YES resolves. Your YES conditional perp wins and settles to the underlying — fine. But now suppose the event resolves NO, or Voids. Your YES conditional perp voids (P&L zero, margin returned). What you are left holding is the underlying-perp leg on its own — the conditional perp that was offsetting it has settled away.
In other words: a conditional-perp-plus-perp "hedge" is a hedge in the branch where the conditional perp wins and a one-sided position in the branch where it loses. When resolution lands on the branch where your conditional perp voids, your remaining perp leg is exposed at full delta.
Resolution and the same-block risk sweep
Resolution settles the family and then the exchange's normal end-of-block risk sweep runs in that same block. If settlement has left you with a now-naked perp position that fails maintenance margin, it can be acted on by the risk engine as early as the same block as resolution, on the post-settlement picture. The risk engine evaluates your account on the post-settlement state in the same block, so size your position to be solvent in that state ahead of time.
What to do
The rule is simple: be solvent in the post-resolution world. Before a family resolves:
- Flatten your conditional position, or
- Fully box it so it's self-contained in every branch — for instance, a position whose payoff is locked at entry across YES, NO, and Void, so that no leg goes naked no matter how the event lands.
A fully matched, symmetric set across the family can have its payoff fixed in every branch and so carries essentially no resolution risk. A partial or one-sided position does not — the exposure shows up at the void/losing branch, where the conditional perp that was offsetting the perp leg has voided.
The conditional legs freeze before the event resolves, and the resolution-plus-settlement sequence is fast and automatic, so this is standard practice: manage the position before the freeze: flatten or fully box ahead of the deadline, not at it.
Quick reference
- Three outcomes: YES, NO, Void. All five books settle atomically. Void is never auto-derived — it's the operator escape hatch and always overrides.
- Winning conditional perp: cash-settles at the underlying's mark, like closing a perp.
- Losing or voided conditional perp: voided — P&L zero, margin returned. Never a settle-to-$0 loss.
- Prediction binaries: win → $1, lose → $0, Void → $0 on both. The binary box (BY + BN ≈ $1) holds through YES/NO and breaks to $0 through a Void.
- Early close issues a $0-entry binary, not cash. Closing a conditional perp before resolution issues a $0-entry prediction-binary position in the firing branch — long on a gain, short on a loss (~$1 of P&L per token; sub-$1 / sub-cent amounts round to zero). That position then resolves with the event ($1/$0), sells on the binary book at the implied probability for cash today (≈ implied probability × tokens), or — if you already hold the opposing binary — realizes cash directly. The binary is the cash-out; there is no separate cash primitive.
- Conditional P&L is universe-internal until resolution — a branch-contingent claim, not withdrawable cash, because cash only exists once the event settles. The $0-entry binary is the tradeable form of that claim.
- Flatten before resolution. A conditional-perp-plus-perp hedge holds in the branch where the conditional perp wins; in the void/losing branch the conditional perp voids and the perp leg is left at full delta, and the end-of-block sweep evaluates your account that same block. Flatten or fully box ahead of the pre-resolution freeze.