Launch app
Launch app

Scalar vs binary prediction markets.

Both kinds of market ask about the same number. One pays you for landing in a box somebody drew, the other pays you for how close you got. Almost everything else follows from that.

How does a binary prediction market pay out?

A binary market turns a number into a yes or no question. Will Bitcoin finish above 55,000. You buy yes at 40 cents, and if it does, your share settles at a dollar. If it does not, your share is worth nothing.

That works cleanly for questions that are yes or no. A number is not one of those questions, so the market has to invent a threshold to make it fit.

Who chooses the range in a prediction market?

Somebody does, and they are not required to explain it. Ask why the wall sits at 55,000 and not 54,500 and there is no answer beyond it being a round number.

One outcome, two walls
You bet Bitcoin finishes above the wall.54,00056,000wall at 54,500$1wall at 55,000$0finishes at 54,900Same outcome, opposite answer, and both walls are round numbers.
Illustrative. Both walls are round numbers and the outcome never moved.

Bitcoin finishes at 54,900, which makes you right against a wall at 54,500 and wrong against a wall at 55,000. Your read never changed, and somebody else's choice of a round number decided the outcome.

There is a fair defence of round walls. Option strikes cluster at them, so attention and liquidity gather there too. That explains why 55,000 gets picked instead of 54,873. It does not explain why 55,001 should be worth a dollar while 54,999 is worth nothing.

What happens if you are almost right?

In a binary market, nothing. You are paid for the box, not the call.

The jump at the wall
$0$154,999pays nothing55,001pays everythingTwo dollars apart in the world, a dollar apart in your account.
Illustrative. Payout is flat on either side and jumps at the threshold.

Finish at 55,001 and you collect everything, finish at 54,999 and you collect nothing, and two dollars is all that separates them. Everywhere else in trading, being nearly right pays nearly as well as being exactly right. A wall removes that, and it does so at the level where the most people have a view.

On a scalar market the payout tracks the finish. Land halfway up the range and you are paid halfway. Miss by a little and you lose a little.

Why do more buckets make trading worse?

The usual repair for one arbitrary wall is more walls. List ten boxes and let people pick. It fixes what you can express and damages how you trade it.

Same orders, cut five ways
One market$100kone book→Five buckets$20k$20k$20k$20k$20kfive books, one fifth as deep eachNobody left. The question was cut up, so the depth was too.
Illustrative. Every bucket is a separate order book.

A bucket is not a view, it is a market, with its own bids, offers, and spread. Split one question five ways and the traders who were meeting in one place now stand in five rooms. Thin books quote wider, so getting in and getting out both cost more.

The quieter cost is the trader who agrees with you, standing in the box next door where the two of you cannot reach each other. A scalar market puts every opinion about the number into one book, so they can.

How is a scalar prediction market settled?

This is the question worth asking any venue. Who decides what happened, and can they change their mind afterwards?

Many markets answer with a vote. Holders decide, a dispute window opens, and a contested outcome can be reopened. That is invisible on small, quiet questions and hard to miss on large, close ones.

Two ways to settle
Settled by voteThe number lands→Holders vote→Dispute window→?Settled by ruleThe number lands→Administrator prints→Published rule→$0.62 to stressOne path ends in a number. The other ends in an argument.
Illustrative. Both paths begin at the same place.

It also decides whether an institution can take the other side. A desk marks a position daily, and marking it means knowing what it pays under every outcome. If the payout can be voted away, that number does not exist, only an estimate of how a vote might go. You cannot model it or put it in a risk report.

A RAVA market settles differently. An independent administrator publishes an index level every day under a public methodology, and at the end of the window settlement reads that level and applies a rule published before trading opened. Nobody votes and nothing is interpreted.

How are the floor and cap set?

A scalar market has bounds of its own, so the fair challenge is what makes them different from a wall. A bucket wall is chosen, usually because it is round, and it decides whether you are paid at all. A RAVA bound is computed from how the number has behaved before, and it decides only how far the scale runs.

Bounds taken from the record
floorcapworsebetterThe bounds sit where the record reaches, not where a round number falls.
Illustrative shape, drawn for teaching and not read off the calibration.

The rule is published and the bounds are fixed before the window opens, so no outcome can be made cheaper or dearer afterwards by moving the scale. The current series runs from a floor of minus 5 percent to a cap of plus 12 percent.

Being near a bound is also nothing like being near a wall. Near a wall, one dollar swings the whole position. Near a bound the payout is already close to its limit, so the last stretch barely moves it.

Which one should you trade?

If the question ends in yes or no, a binary market fits it.

If it ends in a number, the choice comes down to what you are being asked to get right. A bucket market asks you to be right about a box somebody invented. A scalar market asks you to be right about direction and size, which is the view you had in the first place.

That is the trade worth taking. Being directionally right pays in proportion to how right you were, and no round number sits in the middle deciding whether the call counted at all.

The mechanics behind the current series live in the index and settlement, and markets shows what is listed now.