Peer lets you take a position on how a trader does, rather than on what they hold.

FOMO is a social app where people trade in public. You can already follow someone there and watch every position they open. What you cannot do is act on the view you form while watching. Peer is the missing half: it turns a trader into something you can be right or wrong about, with money on it.

A market on Peer is a plain yes-or-no question with a deadline. Will @unipcs be up in 24 hours? You buy the side you believe. If you are right, each share you hold pays out 1 USDG. If you are wrong, it pays nothing. There is no leverage, no liquidation, and nothing to manage in between.

The trader, not the token.

This is the distinction the whole product rests on, so it is worth being blunt about. If you buy a memecoin, you are exposed to that coin. If you take a position on a trader, you are exposed to their judgement across everything they hold, including the coins they have not bought yet.

A trader who is up because they called one thing correctly and a trader who is up because they are consistently early both show up the same way here: as a line that goes up. What that line is made of is theirs to decide and yours to have an opinion about.

Here is every step of a $100.00 ticket, with the real numbers from the live market on @unipcs.

1. 01

   Pick a trader and a question

   Will @unipcs be up over the next 24 hours? The market closes at a fixed time, and the answer is decided by their total account PnL on fomo, not by anybody’s opinion.
2. 02

   Look at the price

   Up is trading at 43¢. That is the market saying it is roughly 43% likely. A price is a probability here, which is the single most useful thing to know about this product.
3. 03

   Buy the side you believe

   $100.00 at 43¢ buys 229.89 shares. There is no fee on the way in.
4. 04

   Wait, or sell early

   Your position is worth what the market will pay for it at any moment. If the odds move your way you can sell before the close and take the difference. Nothing forces you to hold to settlement.
5. 05

   Settlement decides it

   At the close the keeper publishes the trader’s total PnL, and every share on the winning side becomes worth exactly 1 USDG. Losing shares become worth nothing.
6. 06

   Redeem

   229.89 winning shares redeem for $229.89 gross. The protocol takes 2% of the winnings only, which is $2.60, leaving $227.29. That is a 2.27x return on the ticket. If you were wrong, you lost the $100.00 and nothing else.

liquidity          $0.26   10%
flywheel           $2.34   90%
  burn PEER         $1.17   45%
  to holders       $1.17   45%
trader escrow      $0.00   0%

One number per trader: how much money they are up or down on fomo, all time, in dollars. Every market on Peer settles against it.

It is the figure fomo’s own leaderboard ranks people by, and it is the figure the trader is judged by socially, which is exactly why it is worth trading. A market asks whether it is higher at the close than it was at the open. Nothing else about the account matters to a settlement.

The number is signed. Plenty of accounts are down, and a trader at -$40,000 who ends the day at -$25,000 was up over that window. Direction over the window is the whole question; the level it starts from is not.

The reason the underlying is PnL and not the balance in the account.

Suppose a market settled on portfolio value. A trader wires in $10,000 an hour before the close, their balance jumps, and every holder of the up side wins a bet they did not earn. Anybody watching could see that coming and take the free side of it, which is not a market, it is a tap.

Profit and loss is already net of that. Depositing moves the balance and leaves PnL exactly where it was; only a trade that made or lost money moves it. There is no formula here and no correction to apply, which is the point: the property the index had to be constructed to guarantee is one this number has by definition.

A keeper that reads fomo every five minutes and writes down what it saw.

fomo has no developer API and does not publish its traders’ wallet addresses, so the profiles are the entire surface. The source of truth is fomo’s own leaderboard, read by a service that never stops, and the reading is kept as a file rather than a number in a database.

1. 01

   The keeper holds a session

   The leaderboard is behind a login, so the keeper keeps one browser page open and lets the site’s own SDK refresh its token. A read from outside a browser is refused by fomo whatever headers it carries, so this is not a preference: it is the only thing that works.
2. 02

   Every five minutes, a snapshot

   Each read writes one file: the time, the source, and every listed trader’s handle and total PnL. About a kilobyte, 288 a day. This is the evidence, and everything downstream is derived from it.
3. 03

   Published, not just stored

   The keeper serves its own snapshots and settlements read-only. A settlement commits the hash of a file, and the hash of a file nobody can read proves nothing to anyone.

The windows are measured here rather than requested. fomo’s window=24h and window=7d return byte-identical rows, so both are decorative: there is one board carrying one lifetime total, and a 24h market is that total sampled a day apart from our own snapshots.

Neither end of a market is decided by a single reading.

A market’s strike is the trader’s total PnL at the open and its settlement is their total PnL at the close, and both are taken as the median of the three snapshots nearest that moment. Three reads spanning fifteen minutes is what makes a last-second flush worthless: moving one reading moves nothing, because the median ignores it.

strike     = median(3 snapshots nearest opensAt)
settlement = median(3 snapshots nearest closesAt)
outcome    = settlement > strike ? call : put

A snapshot older than 20 minutes at the moment being valued is not eligible, so a keeper that stopped reading cannot have its last good file stretched over a gap it did not observe. If three usable snapshots do not exist, the settlement job refuses to resolve rather than guess, and the market becomes a void. See when a market voids.

One question, asked at two horizons: will this trader be up over the next day, and over the next week? Ten traders, two markets each, twenty live markets. Up is a call on their PnL, Down is a put.

Both settle on the same underlying: the sign of the trader’s PnL over the window, read off the Peer Index. Only the horizon changes, which means the index only has to be right once for every market on the platform to be right.

Further templates are already defined in the protocol: activity markets on whether they open a new position, target-return and range-bound markets shaped by their volatility, streaks, and head to head pairs. They list once the launch pair has volume; nothing about the machinery below changes when they do.

Every v1 market opened at exactly 50c, on purpose.

A market with no trades in it still needs a price, and there are two ways to supply one: compute a probability from the trader’s history, or decline to and let the first traders find it. v1 declines. All twenty markets were listed at 50%, seeded with $10.00 of liquidity each, and moved from there by whoever traded first.

The pricing model is still in packages/core and the contract still stores an opening probability per market, so seeding from a model is a listing-time argument rather than a rewrite. It stays unused until there is enough of our own history to justify a number other than a half.

w        = historyDays / (historyDays + priorDays)
μ        = w · traderDrift + (1 − w) · cohortDrift
scale    = σ_window · √((ν − 2) / ν)
z        = (ln K − (μ·T − σ_window² / 2)) / scale
P_model  = 1 − T_ν(z)

α        = clamp(historyDays / priorDays, 0, 1)
P        = clamp(α · P_model + (1 − α) · 0.5, 0.05, 0.95)

There is no order book. A market maker holds reserves of both outcomes, and the price is the ratio between them.

An order book needs two people who want opposite things at the same moment, which a market on a specific trader closing on a specific day will rarely have. So Peer uses an automated market maker instead: it always has a price, and it always takes the other side.

Buying Up removes Up shares from the reserve and adds to Down, so the price of Up rises. That is the entire mechanism, and it is why a big order costs more per share than a small one.

p_up       = R_down / (R_up + R_down)

sharesOut  = I × (R_up + R_down + I) / (R_other + I)

A winning share is worth 1 USDG, less 2% of the profit. Cheaper shares pay more, because they were less likely to win.

The curve is steep at the left because that is where the unlikely outcomes are. A share bought at 10¢ returns nearly ten times if it lands; a share bought at 90¢ returns about a tenth. Neither is a better deal than the other. They are the same deal at different odds.

winnings = shares − shares × avgPrice
payout   = shares − winnings × fee
multiple = (1 − fee × (1 − p)) / p

Settlement is the median of the three snapshots nearest the close, not a reading at the bell.

If a market settled on a single instant, the person with the most to gain from moving that instant would be the trader themselves. A median over the final quarter of an hour removes that: to move it you would have to hold a manipulated figure for the whole window, in public, against anyone willing to trade against you.

1. 01

   The window closes

   Trading stops at closesAt. The contract will not accept a buy after it and will not accept a resolution before it.
2. 02

   The settlement job reads its own evidence

   It takes the median of the three eligible snapshots nearest the close and compares it to the baseline recorded at the open. Higher settles Call; anything else settles Put. The comparison is one shared function, so the number printed by the dry run is the number that goes on chain.
3. 03

   The outcome is committed with its evidence

   The resolver sends the outcome and keccak256 of the settlement file. The file names both PnL figures, the window, and the snapshots they came from, and it is served publicly, so the hash on chain is checkable by anyone who cares to.
4. 04

   Winners redeem

   Every share on the winning side becomes worth exactly 1 USDG, less the 2% fee on winnings. Losing shares are worth nothing. Redemption is pull, not push: nobody has to run a distribution.

Voiding returns everybody’s money at cost. It is the designed answer to not knowing, and v1 leans on it.

The condition that actually fires in v1 is the first one. Everything the protocol can settle comes from snapshots, so a gap in the snapshots is a gap in the evidence, and a market whose window contains one cannot be resolved from anything.

Three further conditions sit in the contract for the index-based markets this protocol is built toward: the wallet set behind a handle changing, a portfolio too small for percentage moves to mean anything, and a book the price sources cannot value. None of them can fire on a market written on total PnL, and they are named here because they are in the enum rather than because v1 uses them.

You are owed money whether or not you have ever heard of Peer, and you can leave with one signature.

Listing is permissionless, so a market can exist on you without your involvement. The contract carries an escrow that accrues to a hash of your handle — no wallet needed, nothing to opt into — and its share is currently set to 0%, because v1 has nobody to process claims. Turning it back on is a single constant, and anything already accrued is still there.

One signature opts you out entirely, which voids every open market on you and refunds each position at cost.

One card, repeated everywhere, and the token system it is built from.

Peer is a social profile with a fund card underneath it and live odds where the Buy button would be. Every screen in the product is a variation on this component, which is why the leaderboard, the discover grid and the hero all feel like the same object.

Unipcs

@unipcs·373K followers

"If the Lord will, we shall live, and do this, or that." - James 4:15

PnL · 24H2,791.36

+4.9%

24H24H, +4.9%7D7D, +39.6%30D30D, +13.6%ALLALL, +179.1%

Index

2,791.362,791.36

Vol (σ)

119%119%

Win rate

72%72%

Volume 30d

$18.9M$18.9M

↑Call2.27x2.27x↓Put1.75x1.75x

Will @unipcs be up over the next 24 hours?$106K vol

PEER is live on Robinhood Chain. It is not what markets settle in.

Worth separating the two things, because they are easy to confuse. Markets on Peer are priced, traded and settled in USDG, the stablecoin bridged USDC arrives as on Robinhood Chain. You do not need to hold PEER to take a position on anybody.

Everything the contracts are configured with, in one table.

The dozen words this page uses in a specific way.

Total account PnL

How much money a trader is up or down on fomo, all time, in dollars. The underlying every market settles on.

Snapshot

One reading of the leaderboard, written to a file with its time and source. The evidence behind every strike and every settlement.

Keeper

The service that takes those readings every five minutes and serves them publicly.

Baseline

The trader's total PnL at the open, committed on chain as the market's strike. A record, not an input.

Median rule

Both ends of a market are the median of the 3 snapshots nearest that moment, so no single reading decides anything.

Call

The call side: pays if PnL is higher at the close than at the open.

Put

The put side: pays if it is not.

Binary

A contract worth exactly 1 unit of collateral if its condition holds and nothing if it does not.

Price

What one contract costs. Because it pays 1 or 0, the price is also the market's estimate of the probability.

Payout multiple

What a winning ticket returns, net of the redemption fee. The reciprocal of the price, roughly.

FPMM

Fixed-product market maker. Holds reserves of both outcomes and prices from their ratio.

Void

A market stopped early and refunded at cost, because a condition made settling unsafe.