Every decentralized stablecoin ever built has the same skeleton: collateral on one side, debt on the other, and a liquidation engine frantically keeping them balanced.
What if you threw out the liquidation engine entirely?
That is the question Vitalik Buterin posed in a recent ethresear.ch post. The answer he arrives at is elegant: build synthetics on top of options, not debt.
No liquidations. No real-time oracles. No sudden bankruptcy cascades. Just a slower, gentler kind of risk that users can manage themselves.
The Core Problem With Debt-Based Synthetics
To understand why this matters, you have to understand the original problem.
Say you want to create a synthetic asset that tracks the USD/ETH price. You hold ETH in a vault, mint synthetic USD against it, and promise holders they can always redeem.
But ETH is volatile. If the price drops far enough, the collateral backing the synthetic USD becomes insufficient. Someone is holding the bag of negative value.
Traditional systems handle this with forced liquidation. When collateral ratio drops below a threshold, anyone can liquidate the position: send USD, claim the underlying ETH.
This works, but it has a fatal dependency: real-time price oracles.
- Real-time oracles are the hardest to secure
- They require a small set of trusted actors watching continuously
- They cannot incorporate slow, safe mechanisms like prediction-market-backed dispute resolution
- They are vulnerable to flash crashes, manipulation, and timing attacks
Every major DeFi hack involving synthetic assets traces back to this oracle problem.
The Options-Based Alternative: P and N Tokens
Vitalik proposes two token types: P and N.
The setup is simple. At any time, you can split 1 ETH into a (P, N) pair. At any time, you can recombine P + N to get 1 ETH back. The parameters are:
- T: the ticker being tracked (e.g., USD/ETH price)
- S: the strike price
- M: the maturity date
At maturity, the oracle reports the value of T. Call it x. Then:
- P receives min(1, S/x) ETH
- N receives max(0, 1 - S/x) ETH
That is it. P + N always equals 1 ETH. There is no liquidation because there is no debt.
P is the safe side. If you hold P with a deep out-of-the-money strike (e.g., S = 1500 when ETH is trading at 2500), you effectively hold something close to a USD-pegged asset. N is the risky side, held by speculators who want leveraged ETH exposure.
Why This Changes Everything
The comparison is stark:
Liquidation-based synthetics:
- Normal behavior: stable, until suddenly it is not
- Extreme behavior: forced liquidation, cascading failures
- Oracle: must be real-time, inherently weakens security
- Vulnerability: high to temporary price spikes and manipulation
Options-based synthetics:
- Normal behavior: stable, with slow quadratic drift
- Extreme behavior: gradual adjustment, no cliff edges
- Oracle: slow oracles are fine, even weeks-long delays work
- Vulnerability: depends on user rebalancing strategy, user is in control
The key insight: by moving the decision of when to rebalance from the protocol to the user, you eliminate the need for a global real-time oracle. Users can hide which oracle they use, query multiple sources privately, and take the median.
The Quadratic Drift Trade-Off
Options-based synthetics are not perfect. The main cost is what Vitalik calls quadratic drift.
When you hold a P token, your effective USD exposure gradually deviates from a perfect peg as the price moves. This deviation grows quadratically as you approach the strike price. It is not a sudden liquidation, but a slow, predictable leak.
Here is the counterintuitive part: this drift is probably fine.
Fiat currencies themselves move more than 1 to 4 percent per year against each other. Each person is real-world expenditure basket has its own volatility that dwarfs the drift. Even existing algorithmic stablecoins like RAI see their equilibrium rate move by similar magnitudes.
If you view synthetic assets as tools for price stability of future expenses rather than perfect accounting units, quadratic drift of 1 to 4 percent annually is not just acceptable, it is genuinely competitive.
What Can You Build on This?
The P/N primitive is a building block. Here are the tools and products that could be constructed on top:
1. Options-Based Stablecoin
A fully-automated DAO wraps P tokens with conservative strike prices and handles rebalancing. No voting, no AI, pure rule-based rotation. The result: a stablecoin that does not need real-time oracles or liquidation engines.
2. CPI-Tracking Asset
Set T to CPI/ETH instead of USD/ETH. Users get exposure to cost-of-living indexing without fiat dependence. Could track rent, groceries, or energy baskets in any region.
3. Client-Side Rebalancing Daemon
A local agent that monitors P-token positions and rotates them before getting close to the strike. The user chooses their own oracles privately. Transactions are not pre-visible on-chain, reducing MEV exposure.
4. Low-Slippage Rebalancing AMM
The biggest risk to this entire scheme is slippage from repeated rebalancing. A specialized AMM optimized for patient, one-sided rebalancing orders (not instant swaps) could make rotation nearly free. Think batch auctions purpose-built for option migration.
5. Prediction Market Oracle Layer
Since P/N pairs are literally scalar prediction markets, they can share oracle infrastructure with platforms like Seer or Polymarket. One oracle serves both synthetics and prediction markets, strengthening security through shared economic incentives.
6. Commodity Synthetics
T = gold/ETH, oil/ETH, wheat/ETH. Trustless exposure to commodity prices for ETH holders, with no centralized issuer dependency.
7. Yield-Generating N-Token Vault
N tokens represent speculative exposure. A vault that systematically provides liquidity for N tokens or market-makes them could generate consistent yield for depositors, essentially running an on-chain options strategy.
Shared Oracle Infrastructure
One of the most underappreciated aspects of this design: P/N pairs are structurally identical to scalar prediction markets. The same oracle that resolves Will ETH be above $2000 on June 30? also resolves the P/N maturity.
This means synthetic assets and prediction markets can pool their security budget. More economic activity flowing through the same oracle means more incentive for honest reporting, more participants watching for manipulation, and stronger guarantees for everyone.
This is a rare case where composability actually improves security rather than creating contagion risk.
The Slippage Question
Vitalik flags slippage as the largest existential risk to the options-based approach. If users lose 2 percent or more per year to slippage across multiple rebalancing rounds, the system becomes uncompetitive with traditional alternatives.
But there is reason for optimism: users have extremely low time preference for rebalancing. You do not need to rotate today, you can wait three days for favorable market conditions. This transforms rebalancing from expensive instant swaps into patient, one-sided market making.
The right market structure has not been built yet. Whoever builds it captures significant value.
Implications for the Broader DeFi Ecosystem
If options-based synthetics gain traction, several things happen:
- Oracle design shifts: Projects optimize for slow, safe, dispute-capable oracles rather than real-time feeds
- MEV surface shrinks: No protocol-level liquidations means fewer visible transaction opportunities for MEV extraction
- User agency increases: Users choose their rebalancing strategy, oracle sources, and risk tolerance instead of trusting protocol parameters
- Prediction markets grow: Shared oracle infrastructure means prediction markets and synthetics reinforce each other
Conclusion
The debt-based approach to synthetic assets has dominated DeFi for years. It works, but it inherits a fundamental tension: the need for real-time oracles creates an attack surface that cannot be fully closed.
The options-based approach does not eliminate risk. Quadratic drift is real. Slippage is a concern. The accounting is less clean than a perfect 1:1 peg.
But it removes the worst failure modes. No liquidation cascades. No oracle manipulation triggering instant insolvency. No need to trust a small group of real-time data providers.
Instead, you get a system where risk is gradual, predictable, and user-controlled. Where security comes from patience rather than speed. Where the same infrastructure powers both financial synthetics and public goods like prediction markets.
That might be the better trade.
Based on research by Vitalik Buterin, originally published on ethresear.ch. Special thanks to Vladimir Novakovski, Curve developers, and others for feedback and review.