Binary Options vs Prediction Markets vs Perpetuals

A binary option and a prediction market share the same payoff shape: a fixed amount if a stated condition is true at settlement, nothing if it is not. What separates them in practice is the condition and the resolution path. A binary option's condition is a price level at a fixed expiry, decided by reading an oracle at a single instant. A prediction market's condition can be any verifiable event, and it usually resolves through a reporter, a committee or a dispute window. Perpetual swaps and futures are a different instrument entirely — their payoff is linear in the underlying, so the size of the move is what pays you, not merely its direction.

That difference decides which wrapper fits which question. If your view is "BTC finishes this hour above 118,400" — a hypothetical level, used here purely to make the arithmetic concrete — a binary expresses it exactly: maximum loss is the premium, there is no liquidation price to defend, and the contract retires itself on schedule. If your view is "BTC runs 8% this week", a binary discards the part you care about — the magnitude — and a perpetual is the honest tool. What follows is the mechanics of all three, a side-by-side table, a fully worked payoff example with clearly illustrative numbers, and how DeepBook Predict, the on-chain binary options protocol on Sui, implements the binary case.

Three payoff shapes over the same underlying

Strip the branding off and there are only three payoff functions in circulation. Almost everything else — margin rules, fee schedules, UI — is downstream of which one a contract implements.

The binary: a step function

A binary option pays a fixed amount — call it one unit of collateral — if the underlying settles on the specified side of a strike at expiry, and zero otherwise. Plotted against settlement price the payoff is a step: flat at zero, a vertical jump at the strike, flat at the payout beyond it. Everything interesting about the instrument follows from that discontinuity. Because the payoff is floored at 0 and capped at 1, the contract's fair value is bounded to the same interval, and under a risk-neutral measure that value *is* the implied probability of the condition being true. A contract quoted at 0.42 implies roughly 42%, net of spread and carry. Worth keeping straight: that is a risk-neutral probability, reflecting what the market will pay to transfer the risk, and it is not the same object as a real-world forecast.

The prediction market share: the same step, a different question

A prediction market share is economically the same step function. You buy YES or NO somewhere between 0 and 1 and it redeems at 1 or 0 once the event resolves. The differences sit at the end of the lifecycle rather than in the payoff. The underlying is frequently not a price at all — an election, a policy decision, a shipping date — so there is no feed that can be read at a fixed instant, and the outcome has to be assigned by a process with an appeal path.

That process is a real cost, not a formality. A market that will not settle for six months ties up collateral for six months, and a market whose resolution criteria are loosely worded ends up argued in a dispute window rather than decided by a fact. Neither problem exists in a well-specified price binary, where the criterion is "the oracle printed X at time T" and settlement is a read rather than a judgement.

The perpetual or future: a line

A perpetual swap pays the change in the underlying's price multiplied by position size, continuously, with no expiry; a dated future does the same until delivery. The payoff is linear and, on the long side, unbounded. That linearity is the point — it lets you size exposure to conviction, and it makes hedging clean, because a linear payoff cancels a linear risk. It also drags in the machinery that leveraged linear exposure requires: margin, a mark price, funding payments to keep the contract tethered to spot, a liquidation engine, and some backstop for when liquidation is not fast enough. None of that exists in a fully collateralised binary, because there is nothing to liquidate. You already paid the most you can lose.

Binary options vs prediction markets vs perpetuals

The three overlap enough that the vocabulary gets used loosely — "prediction market" in particular is often applied to anything with a yes/no outcome. The differences that actually change how you trade are these.

A worked example

The numbers below are illustrative and chosen to make the arithmetic legible. They are not quotes, not historical fills and not platform data. Fees, spread and slippage are excluded so the payoff structure stays visible.

Underlying

BTC/USD (hypothetical)

Assumed price at open

118,000

Strike

118,400 (about 0.34% above the open)

Side

Above

Expiry

60 minutes from the open

Payout if true

1.00 per contract

Price paid

0.42 per contract

Contracts bought

100

Total premium at risk

42.00

Paying 0.42 for something that pays 1.00 puts breakeven implied probability at 42%. Buy it believing the true chance is 50% and you are being paid for an edge; buy it believing 35% and you are the one supplying the edge. Notice how small the required move is relative to the horizon: a strike a third of a percent above spot sits well inside the range BTC can cover in an hour, so neither outcome is close to settled at the moment of purchase. That is the structural reason a contract specified like this one prices away from the extremes rather than near 0 or 1 — the distance to the strike is small compared with the volatility available before expiry.

Two pairs of rows are worth staring at. A print of 118,401 pays exactly what a print of 121,900 pays — the binary discards magnitude completely. And a print of 118,399 loses exactly what a 5% collapse loses. The instrument compresses an entire distribution into one question, which makes it efficient when the question was genuinely binary and wasteful when it was not. If your edge is in forecasting *how far*, a binary will not pay you for it.

The same 42 in a perpetual

For contrast, put the same 42 of collateral into a 5x long controlling roughly 210 of notional, again ignoring fees and funding. The 3.3% move up returns about 6.90, or 16% on collateral, against 138% for the binary. The 0.34% move up returns about 0.70 — near nothing — against the binary's full 138%. The 5.1% move down loses about 10.70, against the binary's total 42. But an adverse move approaching roughly 20% exhausts the collateral behind 5x leverage — in practice the position is closed before that, because maintenance margin triggers while some collateral remains — and it can happen on a wick that retraces a minute later. The binary cannot be stopped out early. The perpetual can, and frequently is.

What "defined risk" actually buys you

Defined risk is a structural consequence of full collateralisation, not a marketing adjective. Because the maximum payout is known when the position is minted, it can be escrowed up front, and a large amount of derivatives infrastructure simply stops being necessary.

• No liquidation engine. There is no maintenance margin, so there is no price at which the position is closed against your will and no forced-selling cascade to be caught in.
• No funding rate. Nothing has to tether the contract to spot, because the contract is defined by one oracle read at one timestamp rather than by a continuously tracked index.
• No socialised loss. Auto-deleveraging and insurance funds exist to cover shortfalls when a leveraged account blows through bankruptcy. A fully collateralised binary cannot go past bankruptcy — the worst case was funded at entry.
• Position size is the risk number. Capital at risk equals premium paid. There is no second calculation involving margin ratios, and no way for an overnight gap to make the number larger than it was when you went to sleep.

What defined risk does not mean is low risk. For an out-of-the-money binary, losing 100% of the premium is the *modal* outcome — the payout is priced on exactly that basis. And risk near the strike is discontinuous in a way linear instruments never are. In the final minutes before expiry, a near-the-money binary's sensitivity to spot concentrates into a narrow band around the strike: inside that band a small move in the underlying can swing the contract across most of its range, while a few ticks outside it the price barely responds at all. That concentration — not a smooth greek you can hedge away — is why short-dated binaries feel violent even though the maximum loss was fixed the moment you paid.

Expiry mechanics, step by step

The lifecycle of a price binary is short and strictly ordered, which is what makes it implementable on-chain without an off-chain risk system running alongside it.

1. The strike is set

   A round opens and a strike is fixed, typically at or near spot at that moment. From then on the terms are immutable: an underlying, a strike, a side, an expiry. Nothing about the contract is renegotiated while it is alive.

2. The position is opened

   You pay a premium somewhere between zero and the payout and receive the position. On DeepBook Predict that means minting an on-chain position; the claim lives in your own on-chain account rather than as a row in a platform ledger.

3. Time value bleeds

   Between open and expiry the price tracks the market's running estimate of the probability. All else equal — spot unchanged — an out-of-the-money contract decays toward zero and an in-the-money one climbs toward the payout, while a near-the-money one can sit pinned near the middle until the last moments.

4. Expiry reads the oracle

   At the stated time the settlement price is read. Nothing here is discretionary — there is no committee and no appeal window, because the criterion is a number the chain can check rather than a judgement someone has to make.

5. Redemption

   The winning side redeems for the payout and the losing side is worth nothing. On DeepBook Predict's testnet deployment, the settlement asset is dUSDC.

Why binaries fit short-duration crypto price questions

Hourly and daily crypto price markets are close to the ideal use case for a binary payoff, for reasons that are mechanical rather than stylistic.

• The question is already binary. "Above or below this strike at the close of the hour" has two answers. Expressing it through a linear instrument means carrying leverage machinery for a view that never needed leverage.
• Funding may be small at this horizon, but its machinery is not. A perpetual's funding accrues on whatever interval the venue sets, and across a single hour it can contribute very little to P&L while still requiring you to hold margin against a mark price that can gap.
• One oracle read, not a continuous dependency. A binary needs a trustworthy price at exactly one timestamp. A leveraged position needs a trustworthy price every second it is open, because every second is another opportunity to be liquidated on bad data.
• No dispute window can exist at this cadence. A market resolved by reporting and appeals cannot run hourly — the resolution process outlasts the market. Deterministic oracle settlement is not just convenient here, it is the only mechanism fast enough.
• Bounded collateral means bounded quoting. A maker who sells a binary knows the worst case exactly and can escrow it. That is a far simpler book to run than one where the worst case depends on how far the underlying travels.

Where the settlement price comes from

The integrity of a binary rests entirely on its resolution input, which deserves more scrutiny than the payoff does. A step function is a target: if the reference price can be nudged for one second at the expiry timestamp, that second is worth the whole notional to someone. Practical designs mitigate this with the familiar tools — aggregation across many independent publishers, a confidence interval published alongside the price so a consumer can tell a tight quote from a disputed one, and windowed or time-averaged reads instead of a single tick. Oracle resolution and on-chain settlement goes into that in depth.

DeepBook Predict implements the binary case on Sui, alongside DeepBook, Sui's native on-chain central limit order book. Its parameterisation is worth reading closely, because it generalises further than "above or below".

Contract definition

An expiry plus a strike expressed as a price tick

One-sided contracts

Sentinel tick values express "above X" or "below X"

Range contracts

A bounded pair of ticks

Settlement

Against the protocol's on-chain oracle price at expiry; the holder redeems for the payout

Settlement asset (testnet)

dUSDC

Pricing inputs

Spot, forward and volatility-surface feeds published on Sui, plus Pyth price feeds for the underlying

Custody

Positions minted, held and redeemed on-chain — no off-chain matching engine holds the book

The volatility-surface input is what makes this an options protocol rather than a betting venue with a price feed attached. A binary's fair value is not a guess about direction; it is a function of spot, the forward, time to expiry and implied volatility at the relevant strike. Expressing the strike as a price tick rather than a raw number is what lets one contract type cover every case in the spec above: a sentinel tick means "everything on one side of here", a bounded pair means "between these two", and the settlement logic does not have to branch into separate instruments. Publishing the pricing inputs on-chain is what allows a quote to be reproduced by anyone rather than asserted by an operator. What DeepBook Predict is covers the protocol from first principles.

How this maps onto Predict Bay

The markets on Predict Bay are short-duration price markets on BTC, ETH, SOL, SUI, DOGE and XRP, running as hourly and daily rounds — tradeable in practice mode today, for the reasons set out immediately below. A strike is set at the open, you pick above or below, and the market resolves against an on-chain price feed: the binary structure described above, applied to the assets and horizons where it works best. The BTC market page shows the shape of one.

With EVM off, the mode switch on a market page renders two options: Practice and Sui. Signing in is Google via zkLogin, so there is no seed phrase to store or lose, and transactions can be sponsored, so holding SUI for gas is not a prerequisite; connecting a Sui wallet directly is also supported. How it works walks the flow end to end, and the FAQ covers account and settlement questions.

Choosing between the three

• Use a binary when the question has a natural threshold and a natural deadline, when you want loss capped at the ticket price with no margin to manage, and when you accept that a near miss pays the same as a rout.
• Use a prediction market when the question concerns an event rather than a price, and you can tolerate a slower resolution path and collateral tied up until it completes.
• Use a perpetual or future when magnitude is the point — hedging a spot book, sizing to conviction, holding a view over an undefined horizon — and you are equipped to manage margin, funding and liquidation risk continuously.

None of the three dominates the others. They price different objects: a binary prices a probability, a perpetual prices an asset, and a prediction market prices a claim about the world. Picking the wrong wrapper is one of the more common ways a correct view still loses money.