Introduction
Unrig is a prediction market protocol built on Ethereum. It enables counterparties to emerge naturally without any liquidity provision, and eliminates any room for profitable manipulation within its trading mechanism. The first removes the fundamental constraint that prediction markets have lived with since their inception, turning them from a product bounded by liquidity into general-purpose infrastructure for all of humanity’s judgments about the future. The second proves, for the first time in financial history, that a continuously traded market can make manipulation unprofitable through its mathematical properties alone, without relying on any sustained effort from external regulators.
Consider the simplest possible prediction market: those who predicted correctly take the funds of those who predicted incorrectly, in proportion to what they staked. In practice, such a model cannot work at all. Bets placed at different times face very different levels of risk, and if everyone receives the same return no matter when they enter, everyone will wait until the outcome is certain before betting. No counterparties form, and no market can exist. Existing designs solve this with arbitrage driven by price movements, but that requires a market to keep generating enough capital inflow and trading activity — and so the limits of liquidity become the limits of where prediction markets can exist. Unrig does not take that route. It directly measures the real risk each participant bears and makes returns precisely commensurate with risk. Even if a market has only a handful of participants from start to finish and its price barely moves, it still works normally. As a result, on any topic and at any time, a prediction market exists as soon as someone is willing to act.
Manipulation has existed for as long as financial markets have, because every form of it shares the same driver: the manipulator can profit from it. Existing anti-manipulation efforts focus on detecting and blocking specific techniques, but they run into a fundamental dilemma: from the outside, manipulation and normal trading look the same. A large buy followed immediately by a sale could be a pump-and-dump, or simply a change of mind. The outward behavior is identical; only the motive differs, and no system can observe motive. As long as manipulation can be profitable, it will continue, however advanced anti-manipulation tools become. Unrig does not try to block specific techniques. Instead, its mathematical structure removes the possibility of profit itself: each participant’s return is calculated strictly by formula from the real risk they bear, manipulators cannot create any return for themselves beyond what the formula gives them, and the cost of carrying out manipulation is strictly positive. A manipulator can still take any action, but no action will make them a profit. This is not because manipulation is blocked, but because no one has any reason to do it.
After a market expires, who won and who lost must be determined accurately. This is the premise everything rests on: if it fails, everything collapses. Once the liquidity constraint is removed and the number of markets is no longer capped, outcome determination has to keep up with that scale. Unrig has redesigned outcome determination from the ground up as well — built for an unlimited number of markets, and more reliable than any existing approach.
This paper sets out in full the mathematical structure and logic of risk pricing, proves rigorously that manipulators cannot profit, and describes in detail the settlement process, the market creation rules, the outcome determination system, the Community Sovereign Voting mechanism, the economics of the $UNRIG token, the referral distribution network, the ethical boundaries of markets, and contract security and permissions.
Chapter 1: Counterparty Formation
For counterparties to form, someone has to be willing to bet first.
People will act first if the risk of acting first earns a commensurate return. But for that return to reach those who acted first, two further conditions must hold: late capital cannot touch the returns of those who acted first, and losers cannot take winners’ money.
Unrig has designed two mechanisms for this: the risk duration integral (T_acc), and continuous integral pricing for early exits. Both rely on the same number: the option’s market probability p, the share of all capital in the market that is staked on that option. The share staked on the other side, 1 − p, is the probability that you lose your principal at this moment — the risk you bear. T_acc adds up that risk second by second, and returns are distributed according to the total: every second of risk borne by those who acted first is already recorded, and even if the outcome is already determined before the market expires, late capital cannot touch it. Exits settle at p: for each unit of principal you exit, you take back only p, and the lower the probability, the heavier the penalty. Losers cannot take winners’ money.
The rest of this chapter sets out the two mechanisms and shows why manipulators cannot profit within this structure.
1.1 Risk Duration Integral (T_acc)
Each option maintains a variable, T_acc, which grows at the following rate:
T_acc increment per second = (1 − p) ÷ W
W is the total principal of all current positions on the option, M is the total current principal across all options in the market, and p = W ÷ M is the option’s market probability. 1 − p is the share of capital staked on the other side, where the other side means every other option.
Every second, a position accumulates allocation weight equal to the risk it bears. The numerator, 1 − p, is the risk, and it is also the weight its side issues each second. The denominator, W, determines how that weight is divided: the weight issued in a given second is shared among all positions on that side in proportion to principal, so the more principal there is to share it, the less each unit of principal receives.
When you enter, the current value of T_acc is recorded as your entry snapshot. When the market expires, the system locks the final value of T_acc for each option. The final value minus your entry snapshot is the risk duration integral you have accumulated. Your allocation weight is:
Allocation Weight = Principal × Risk Duration Integral
Consider the case where the outcome is already determined before the market expires but betting is still open. Capital that rushes into the winning option at this point is betting on a known outcome. Even so, it cannot touch the weight that those who acted first have already accumulated. The weight for every past second is already recorded; new capital can only accumulate from the moment it arrives, and no amount of money can buy seconds that have already passed. Beyond that, the inflow itself pushes the weight issued in each second after its arrival toward zero: it drives its own side’s p toward 1, so 1 − p approaches zero, and it adds to W, so more principal now shares that weight. However large the inflow, the share it can obtain as a whole is capped, and that share shrinks toward zero as the inflow keeps growing.
Allocation weight can only be determined when the market expires. Your entry snapshot is locked when you bet, but how fast T_acc grows while you hold depends on every participant entering and exiting, on your side and on the other side.
1.2 Early Exit and Penalty Pricing
Before the market expires, any position can be exited early at any time. For each unit of principal exited, you take back p and leave 1 − p behind. By the market’s judgment at that moment, the unit stays with you with probability p and goes to the other side with probability 1 − p; an exit settles only the part that is yours. The 1 − p left behind is the penalty, and it goes to all winners.
Unrig applies continuous integral pricing to exit penalties: the exit amount is treated as infinitely many infinitesimal elements, each priced at its own instantaneous probability. Every portion exited lowers the option’s probability, so exiting the rest costs more.
Let the exit amount be h; before the exit, let the option’s principal be W, the market’s total principal be M, and the other side’s principal B = M − W. The instantaneous probability after s has been exited is:
p(s) = (W − s) ÷ (M − s)
The net proceeds from each infinitesimal element ds are p(s) · ds. Integrating from 0 to h:
Net Proceeds = h − B · ln(M ÷ (M − h))
Total Penalty = B · ln(M ÷ (M − h))
Path independence. The total penalty does not depend on how an exit is split. Split h into two exits, one after the other: when the second begins, W and M have each been reduced by the amount of the first while B is unchanged, so its integration interval picks up exactly where the first left off, and the two together equal a single integral from 0 to h. Exit all at once, in ten installments, or from ten different addresses — the total penalty is the same.
Loser lock-in. As an option’s probability approaches zero, the penalty rate approaches 100%.
1.3 Why Manipulators Cannot Profit
The risk pricing mechanism and the exit penalty mechanism together ensure that the return on every unit of capital is determined strictly by the risk it bears, and the cost of every exit is determined strictly by the state of the market. Both take only objective variables as inputs; neither identifies participants nor judges intent.
In this system, every position held by every participant has only two possible endings: early exit, or holding to settlement. An early exit means paying a penalty, so the net proceeds are strictly less than what was put in — a certain loss. Holding to settlement means the return is determined entirely by whether the prediction was correct and how much risk was borne — that is honest participation. There is no third path. Take the most typical manipulation from traditional markets, as it would play out here: buying heavily while the probability is low, then pushing the probability up to draw in followers. The manipulator, too, must eventually leave through one of these two exits, and neither offers an excess return.
These rules do not depend on detecting manipulation, because manipulation and normal behavior cannot be told apart from the outside. Nor do they depend on oversight or enforcement by any central authority: every rule is written into immutable code on the blockchain and enforced automatically, equally and without exception for all participants, and no one can open a back door for anyone. Within this structure, manipulation is unprofitable. This is not because it is blocked, but because no one has any reason to do it.
Chapter 2: Settlement
When the market expires and the outcome is known, settlement begins. Allocation weights accumulate in real time while the market is open and are locked the moment it expires; once the outcome is known, how much each person receives is fully determined.
Settlement follows a strict flow of funds: total up the fund pool, return all of the winning side’s principal first, treat the remainder as the profit pool, take a 15% protocol fee from the profit pool, and finally distribute the remaining profit to the winning side by allocation weight. The stages run in a fixed order that cannot be reversed, and funds are never mixed between stages. The winning side’s principal is fully set aside before any fee is taken — anyone who predicted correctly and held to settlement cannot lose money by participating.
The full amount of every bet goes into the option’s pool; no fee is deducted on entry. When the market expires, the principal still remaining in each option, plus the total penalties paid by all early exiters while the market was open, constitutes the total fund pool — all of the funds available for distribution.
Total Fund Pool = sum of all options' current remaining principal + Penalty Pool
The first step of settlement is to return the winning side’s principal. Each winner gets back the full current principal of their position in the winning option. Principal in the losing options is not returned; together with the penalty pool, it makes up the profit pool.
Profit Pool = Total Fund Pool − all winning-side principal
Protocol Fee = Profit Pool × 15%
The protocol fee is split among four parties in fixed proportions: creator 30%, staking pool 40%, Protocol Treasury 25%, Founding Team 5%. Where a market has promoters, the promoter receives 15% of the protocol fee and 5% of the $UNRIG minted, in proportion to the promoter weight ratio defined in Chapter 7, and the creator’s corresponding shares are reduced accordingly.
After the protocol fee is taken, the remaining profit is distributed among all holders of the winning option by allocation weight.
Distributable Profit = Profit Pool − Protocol Fee
Individual Profit Distribution = individual allocation weight ÷ total allocation weight of all winners × Distributable Profit
Allocation weight was fully defined in Chapter 1 as the product of principal and the risk duration integral. At settlement, this weight is used directly to distribute profit. Each winner’s net payout is their principal repayment plus their profit distribution.
Individual Net Payout = Principal Repayment + Profit Distribution
Each user can claim only once per market. Everything owed remains claimable indefinitely.