Skip to content

Vault strategies — simulated on real data

idea bank #1 + #2 · live replays

The track brief requires "proper simulation results" for vault strategies. Both simulations below are recomputed from this bot's own recorded surfaces, on-chain LP events, and oracle settlements on every page load — same engine the judges can curl.

One method, two verdicts. The range ladder (auto-mint N adjacent bands around ATM each expiry) is viable — but only at the right width, and the data picks it: σ/2 rungs. The PLP + hedge vault (supply the house side, buy crash insurance) is not viable on today's surface — PLP is the counterparty to the very miscalibration our other strategies harvest, and near-tail insurance costs ~2× its realized payout. Publishing the NO with its numbers is the point: it's the same smile-shape diagnosis, seen from the house's side of the table.

  • Execution exists for both: svx mint-ladder mints the winning σ/2 ladder live via predict::mint_range; svx supply-plp takes a house-side position via predict::supply.
  • Range caveat: ranges have no permissionless redeem — the operator key redeems after settlement (redeem_range).
  • Why σ/2 wins: the calibration exhibit (landing page) shows the surface underprices the center of the distribution. Narrow ladders concentrate capital exactly there; wide ladders donate it back through overpriced wings.

Range ladder — strike-width policy shoot-out

5 rungs around ATM per settled oracle, priced off the surface the vault would have seen, 2% fee. GET /range-sim.

Live replay window: this table recomputes on the settled oracles the running bot still retains full surface data for — few since the feed froze July 12. The research verdict (half-sigma rungs, +10.1%) is the archived 104-oracle replay in the repo's backtest report; this card exists to show the method runs live, not to restate that result.

Loading…

PLP + tail hedge
verdict: not viable — here's why

PLP APY realized from on-chain supply/withdraw events; crash insurance priced per oracle cycle off recorded surfaces, netted against realized crash payouts. GET /plp-sim.

Loading…

Three-protocol margin loop
simulated · live when Predict ships mainnet

Borrow dUSDC on deepbook_margin against an iron_bank USDsui share, deploy into the favored-side strategies, repay from settlements. Strategy leg = this bot's real settled trades; borrow APR is an explicit assumption (no public rate feed). GET /margin-loop.

Fills in as favored-side trades settle on this bot.
CONNECTINGtestnet