Skip to content

INVESTOR_ONE_PAGER.md

Latest commit

 

History

History
107 lines (68 loc) · 3.2 KB

INVESTOR_ONE_PAGER.md

File metadata and controls

107 lines (68 loc) · 3.2 KB

Investor One-Pager

Project

Self-Driving Yield Engine is an autonomous, non-custodial yield vault on BNB Chain.

It rotates capital across:

  • Aster ALP for volatility-linked carry.
  • Pancake V2 LP for fee income.
  • 1001x short hedge for LP base-exposure control.

Core Thesis

[Calm Market]
   |
   v
[LP fees matter more] ---> [Keep more LP exposure]

[Storm Market]
   |
   v
[ALP carry + volatility capture matter more] ---> [Shift toward ALP]
  • Pure LP is vulnerable to impermanent loss.
  • Pure ALP is a concentration bet.
  • The target product shape is a middle path: adaptive allocation plus hedge-aware controls.

Why It Is More Investable Now

  • More Accurate NAV: hedge account value is included in vault accounting.
  • Fairer Share Pricing: virtual shares plus TWAP-vs-spot guards reduce manipulation surface.
  • Cleaner Incentives: no-op cycles no longer farm gas-only bounty.
  • Safer Control Logic: hysteresis, partial hedge close, and ONLY_UNWIND improve defensive behavior.
  • Better Evidence: the repo now includes invariants, adversarial tests, static-analysis triage, and CI.

Assurance Snapshot

This repo is not presented as “fully formally verified production software”.

It is presented as a strategy system with a growing assurance stack:

  • 54/54 contract tests passing locally.
  • 5/5 invariant checks passing.
  • 10/10 symbolic formal properties passing via Halmos.
  • adversarial tests covering dependency and unwind edge cases.
  • Slither reduced to 1 remaining finding family focused on flash-callback event ordering.

Manual callback review: docs/PANCAKECALL_AUDIT.md.

  • optional BNB Chain fork checks for live integration readability.

For the full proof index, see docs/ASSURANCE.md.

Research Snapshot (90d model)

As of 2026-03-08 using trailing 90d CoinGecko BTC data.

Assumptions:

  • TVL = $100k
  • BTC path = CoinGecko daily data
  • scenarios = baseline, stress, funding_adverse, liquidity_crunch, gas_spike
  • comparison = dynamic vs fixed_normal vs pure_alp vs pure_lp

Baseline

  • Dynamic CAGR: 15.09%
  • Dynamic cumulative return: 3.49%
  • Dynamic max drawdown: -0.06%
  • Fixed NORMAL CAGR: 13.61%
  • Pure LP CAGR: -1.60%

Stress

  • Dynamic CAGR: 10.93%
  • Dynamic cumulative return: 2.56%
  • Dynamic max drawdown: -0.17%
  • Fixed NORMAL CAGR: 9.30%
  • Pure LP CAGR: -11.27%

Interpretation

  • The dynamic strategy remains positive in the model under both baseline and stress assumptions.
  • Pure LP remains the cleanest negative control: it shows why IL-aware risk management matters.
  • Pure ALP can outperform in some windows, but that is concentration risk, not the intended product shape.
  • The stronger investor story is now “adaptive strategy + clearer assurance”, not just “higher modeled return”.

Boundaries

  • This is still a research model, not realized production performance.
  • ALP carry and hedge behavior are modeled, not full live liquidation-path replay.
  • The remaining static-analysis hotspot is still the flash callback path, which is exactly where manual audit attention should stay focused.

Visual Asset

  • Embedded README screenshot asset: docs/assets/investor-one-pager.svg