What this design changes

Four common failure modes of on-chain options, and where Split stands on each.

1. No maker, no market

Order book and RFQ venues need makers, and makers need flow. Split's pools are the writer: a pool funded with collateral and a signed quote is a two-way market on day one. The trade-off is that the pool carries the risk and its size limits the largest ticket.

2. Oracle manipulation

Oracle-margined venues read prices to set collateral and trigger liquidations. Split doesn't.

• Physical vaults read no price at settlement. The holder decides to exercise.
• Cash-settled bands read one Chainlink price at settlement, with staleness and round checks. The price splits a pot that is already funded; it cannot make the pool owe more.
• The pricer reads the same feed continuously to quote, and halts when it diverges from a reference by more than 2%.

3. Under-collateralization

Partial collateral needs a liquidation engine, which breaks down in a crash. Split locks the full payout per position, so there is nothing to liquidate and no bad debt.

4. Capital efficiency

Full collateral is expensive. Split narrows it two ways: cash-settled bands lock only the width, not the underlying, and a matched Up and Down on one band share one width. It does not try to match perps on notional per dollar.

Limits

• Depth. A pool can only sell what it can pay, so the largest ticket at a multiple is about free collateral ÷ multiple. Exit before expiry is at the pool's bid.