Prediction Perps - a new asset class

The idea

What if an asset could continually pay a return that depends on something happening in the world right now?

A token whose value depends on the weather.

A token whose value depends on the league position of a team.

A perpetual asset, with no expiry, whose return depends on real world phenomena.

We call these assets Prediction Perp Tokens (PPT).

1. A new kind of asset

Investments have historically paid either a fixed return or a dividend that depends on how much profit a business makes. Prediction Perps introduce a new type of return: one that depends on what is happening in the world right now.

That return can follow the weather (temperature, rainfall), the league position of a sports team, the approval rating of a political party, or many other phenomena.

A position can be held for a day, a season or ten years. It can be a view on where something is heading, a way of backing a side, or a hedge against a phenomenon.

2. What today's markets can't give you

There are already ways to take a position on the weather or on a football team. These are all expiring positions, on specific questions with specific outcomes.

Weather

Weather derivatives trade on traditional exchanges, but they are hard to access, thinly traded and come in only a handful of contracts, mostly short-term binaries, a straight over-under, or plain options. Either way, the contract runs to a date and then it is gone.

Football

Football has betting markets, which have existed for decades. Almost all of them are short-term as well: a bet on one match, or on a season, that settles and is over.

Say I think Everton are going to improve. I'm not sure how far, only that they are heading up the table.

You might say my opinion is Long Everton.

To express this opinion I have to work out a whole set of bets, on matches, on top ten, top six, a European place, and form a specific opinion on every one of them. I have to put a date on it too, this match or this season, when all I believe is that they are getting better.

A PPT makes that literal.

Long Everton becomes one simple asset to buy and hold.

One asset, every time horizon

When people trade a normal asset in a market, each trader arrives with a different question, a different time horizon and different priorities. All of those come together in one place, and that is what makes a long-term price, with liquidity and movement. A trader can take a position on how an asset moves over the next 24 hours, week, month or year, without there being markets dedicated to each time horizon.

Expiring contracts split all of that up, into one market per question and per date. A Prediction Perp brings it back together: every motivation and every time horizon meets in one asset.

3. How it works: the distribution rate

The three requirements

To be that one asset, a Prediction Perp has to satisfy three requirements.

1. No expiry. What it tracks goes on indefinitely, so the asset does too.
2. A link to the world. Its return has to follow a public measurement of the phenomenon.
3. Trust without a middleman. It operates by public rules that enforce themselves, with no one in charge to trust.

From requirements to design

Putting the system on a blockchain is the most straightforward way to satisfy requirements 1 and 3. The interesting part of the design is requirement 2: how to link the thing to the real world.

Our inspiration for requirement 2 comes from traditional perps (perpetual swaps).

The funding rate in traditional perps

A traditional perp follows the spot price of an asset, with no expiry date, and uses a funding rate to keep the perp's price at the spot price. When the perp trades above spot, the long side pays the short side. When it trades below, the short side pays the long side.

A funding rate runs into two obvious difficulties with a Prediction Perp.

1. There is no asset with a spot price to tie the PPT's price to.
2. A funding rate breaks the asset's lack of expiry. In a traditional perp, a holder who can't afford the rate is liquidated, so the position does come to an end. That breaks the first requirement of a Prediction Perp.

However, Prediction Perps can still borrow from the idea to get their link to reality.

The distribution rate in Prediction Perps

The distribution rate is the key mechanic of a Prediction Perp. It is what links the asset to the world.

Every full set of PPTs is backed by $1. We call that $1 the underlying asset. It is deposited to earn yield, and what it earns is paid out to the PPTs.

A traditional perp has a funding rate: a payment between traders that pulls its price towards spot. A Prediction Perp has a distribution rate: a payment to holders, taken from what the underlying asset earns and distributed to the PPTs according to the measurement of the associated phenomenon.

Example one: rain

The first thing to do is define the phenomenon, and then define the positions as a closed set.

The simplest example is a binary Prediction Perp on whether it is a rainy day in London. Made specific: did a designated Met Office weather station in London record rainfall today, at 0.2mm or more?

The full set of positions is then simply {Rain, No Rain}.

The asset now has to be linked to what is going on in the world. We can do that with something like a hybrid of a funding rate and a bet, crossed with DeFi yield.

Just as in prediction markets that follow the Polymarket format, where a full set of tokens is backed by $1, we back each {Rain, No Rain} token pair with $1. That $1 is the underlying asset. It goes in when a pair is created and comes out only when a pair is handed back, so every pair in existence always has exactly $1 behind it. The underlying asset is deposited to earn yield, and what it earns each day is distributed to either Rain or No Rain, depending on that day's outcome. That payment is the distribution, and a token's distribution rate is its share of it.

The fair price of a Rain token would then be the discounted value of the distributions it is expected to receive. Its value is a prediction of how wet it will be in the future.

The price of these tokens would ebb and flow with the seasons and with the forecasts.

Example two: beyond binary

We can extend the distribution beyond a simple binary. For example, we might have a token based on hours of sunshine, with the pair being {Sunshine, No Sunshine}. This pair wouldn't simply pay the whole distribution to one token or the other, but would pay proportionally, perhaps based on the % of daylight hours which were sunny or not sunny.

Example three: league tables

As we extend the idea out we can move to multi-runner systems such as league tables.

For example, we can take the Premier League, with {Man City, Arsenal, Liverpool ...} being the full set of positions. Each $1 of underlying asset then supports one token of each of the teams. The distribution can then be divided by the current league position of each team, with holders of the team in first receiving the lion's share and the team in 20th receiving the crumbs.

In short

The distribution rate does the job a funding rate does in a traditional perp, except that it is only paid to holders and never by them. Nobody can be priced out of a position, so all positions can be held forever. A token's yield is what it is paid in a year divided by its price.

What follows from the design

Liquidity only has to be funded once. A market maker for a prediction market has to be funded again for every new event, because every event ends. A Prediction Perp never ends, so its market maker is funded once and keeps pricing every position from then on. It ends up with a standing pool and liquidity providers, the way any long-lived asset does.

A Prediction Perp needs an oracle. Each day's result has to get on chain: whether it rained, where each club sits in the table.

4. The First Prediction Perp

Football

TopFlight is the first Prediction Perp. It is a Premier League market.

Every club has a token. Each token is paid every day according to where its club sits in the table, and it can be bought, sold or simply held.

We follow Polymarket and use UMA's optimistic oracle, so results reach the chain without centralised infrastructure.

The first, not the last

The same shape works for anything that carries on indefinitely, gets measured regularly and can be divided into a closed set. Temperature can be split into bands, with each day's distribution paid to the band the reading lands in. Voting intention is another possibility. So are popularity indexes, and more.

The first one is ready to try. TopFlight is in beta on testnet, at TopFlight.fun. Our article TopFlight: the football stock market goes through how it works in full, and answers what a trader asks next: where the price comes from, what a token is paid, what happens to a relegated club.