deepskew
ManualOverview

The legibility layer for DeepBook Predict

DeepBook Predict puts options-style volatility trading on-chain: an oracle publishes a volatility surface, a liquidity vault underwrites every position, and traders mint and redeem against it block by block. The information is on-chain. The meaning is not. deepskew reads it, makes it legible, and lets you act on it.

A traditional options desk keeps a wall of tools open for exactly this: a surface viewer, an arbitrage monitor, a risk engine, a blotter. On-chain volatility has none of it. The data is public and the protocol is permissionless, but you still have to decode fixed-point parameters, run the vol math yourself, and rebuild state from a stream of events before you can answer a basic question like is this surface fair or can the vault take this trade.

deepskew closes that gap. It does the decoding, the math, and the reconstruction for you, refuses to show a number without showing what it means, and where a read leads to an action it closes the loop on-chain: trade the surface you just read (mint and redeem binaries against your own PredictManager) and supply or withdraw the PLP vault you just stress-tested.

deepskew is one persistent desk with seven routed views. The market selection, the wallet, and the live query cache live in the shell, so moving between views keeps the data warm.

Desk
/

The single-viewport instrument: the 3D vol surface, the smile with its arbitrage verdict, oracle state, the vault, and a live tape of mints and redeems.

Vol Analytics
/vol

The volatility desk in depth: the smile across log-moneyness, skew and term structure, the risk-neutral density, and smile dynamics over time.

Flow & Edge
/flow

Where the vault's edge is made or lost: every fill scored against its model-fair price, settlement calibration, whale flow, range-product flow, and a connect-gated ticket to mint and redeem binaries on-chain.

Vault
/risk

The vault is the counterparty to every position. Per-expiry exposure, a 5-sigma breach point, exit capacity, a GREEN/AMBER/RED grade, and the supply/withdraw funnel.

Managers
/managers

The desk leaderboard: every trading account ranked by volume with realized and unrealized PnL, drill-through to a per-desk page.

Ops / Health
/ops

One read on whether the market is trustworthy right now: oracle-feed freshness and indexer lag against the chain head.

Cross-Venue
/cross-venue

On-chain volatility in context: Predict vs Deribit DVOL, the vol-risk-premium, and the DeepBook stack composability loop.

VerifyCheck it in 60 seconds
Every figure is computed from current on-chain state, and every figure is checkable. Open the desk, watch the cerulean pulse on each oracle tick, click any settlement to see it on Suiscan, and read the formula behind any number in the glossary. The full path is in Verify It’s Real.
  • Read the surface. The full BTC vol surface in 3D, the current smile with its butterfly and calendar arbitrage verdict, and the risk-neutral density it prices.
  • Read the vault. The PLP vault repriced out to five sigma, per-expiry concentration, exit capacity against the withdrawal limiter, and one graded verdict on the Vault view.
  • Trade it. Mint and redeem binary positions against your own on-chain PredictManager from a connect-gated ticket that leads with the model-fair N(d2), with a real predict::mint / predict::redeem.
  • Supply the vault. Supply or withdraw dUSDC to the PLP vault on-chain, with a real predict::supply / predict::withdraw, clamped to the live limiter budget and free liquidity.
  • Reads and transacts on Sui Testnet today.
  • Follows Predict to Mainnet at launch: the read and write logic is network-agnostic, so it is a network and market-resolution change, not a rewrite.
  • Read-only to read-write is done: the desk now calls the contract, not just reads it.
  • Next: settled-redeem keepers and deeper composability across deepbook_margin and DeepBook spot.