Reference

Evidence and limits

Every number on this site was produced by a test in the repository or read off the live chain. Here is which, and here is what none of it claims.

The evidence

• The math is right

  5 references, ±$0.001

  Black-Scholes in fixed point is checked against double-precision references at five spot-strike-tenor-volatility points, all within a tenth of a cent, and put-call parity is fuzzed across 256 random markets. The CDF polynomial is Horner from the highest coefficient down; folding it the other way costs three percent of probability, which is why the references exist.

• The volatility is checked against the prices

  18 feeds, 5,300 rounds, spread 3.8%

  Not against a quoted options market, which prices the future and charges for the risk of doing so, but against the realized volatility of the very same rounds. SpaceX comes out at 69% from its prices and 70% from its timestamps. Across all eighteen feeds the ratio averages 1.000 with a spread of 3.8%, and the worst asset sits at 0.918. The estimator anyone would reach for was off by more than double on seventeen of them.

• The lifecycle closes

  61 tests, 0 failures

  Write, quote, buy, settle, in the money and out, sold and unsold, paused and live, settled at the right round and refused at the wrong one, plus deposit, pool, write, collect and claim through the vaults. Every gas figure quoted on this site is printed by one of these tests.

• The chain figures are measured

  not quoted from docs

  Block time is timed across twenty thousand blocks, fees are read out of receipts, feed staleness out of round data, and pool depth out of token balances. The scripts are in the repository and run against the public endpoint with no key.

What measuring killed

Two designs died to measurements before this one was built, and both deaths are kept. A pool-anchored oracle died when the feed turned out to lag the pool by six hours while staying within 0.08% of it: the pool needs no correcting. A dividend-arbitrage theory died when the dividend, 0.0566%, came in under the pool fee, 0.3%: there was nothing to take. A market that reads pools was therefore never built, which is also why pushing one, at a measured cost of $291 per percent on the thinnest, buys an attacker nothing here.

Limits

• Seller liquidity

  addressed, unproven

  Every options market lives or dies on people willing to write. The vaults are the answer to the part that is a mechanism: depositors pool shares so nobody funds a whole contract alone. Whether people deposit is a different question, and it is not answered yet. The vaults are live and empty.

• One share per contract

  $140 to $930

  Writing escrows a whole share, so the minimum ticket is the share price. Fractional writing is future work; the current design buys its safety with this bluntness.

• Error grows away from the money

  9% becomes 21%

  A volatility error does not travel to an option price one for one. At the money a 9% error is a 9% price error; ten percent out it is 21%, and twenty percent out it is 37%, because an out-of-the-money option is almost entirely a bet on movement. The markup on each market, a public number on the contract, is what stands between a writer and selling too cheap.

• Realized is not implied

  expected, not error

  This estimator measures volatility that has been; an option is priced on volatility to come. Implied volatility sits above realized for exactly that reason, so a gap between this and a quoted options market is the risk premium rather than a mistake. It is also why the markup is not optional.