Quantis AI Trading
An autonomous AI agent trades a US-stock paper account with four sleeves — a volatility-targeted leveraged index core held overnight, a mechanical sector-rotation sleeve, an intraday opening-range sleeve, and a conditional protective-put hedge — every position wrapped in hard survival mechanics the LLM cannot override.
Personal engineering project · Paper trading only · Not investment advice
A Sharpe-stacker: several uncorrelated, evidence-backed return streams under one risk engine, instead of one prediction bet. An LLM runs the routines, but every sleeve is a formula — the core's leverage is target ÷ realised volatility (capped 2×, held overnight), the sector sleeve is pure 6-month momentum, the intraday sleeve is a zero-LLM breakout script, and the put hedge switches on only when the book is actually levered. The vol-target is the engine; the drawdown ladder, VIX kill-switch, margin-distance gate and hedge are the defence. Survival is rule one — a wipeout can't compound.
How it works
Size to volatility
The pre-market routine measures the core's realised volatility and sets its leverage by formula: target ÷ realised vol, capped at 2.0×. Calm tape earns more leverage; turbulent tape automatically gets less — before VIX even spikes. Below the 200-day moving average the exposure is halved. On the first session of each month it also re-ranks all 11 sector ETFs by 6-month momentum and rotates the sector sleeve into the top three. The numbers are calculations, not vibes.
Deploy the sleeves
The leveraged index core is sized to the vol-target, gets a catastrophic backstop stop, and — new in v4.0 — is HELD overnight: 21 years of data show index returns accrue overnight, and the old habit of de-levering into every close gave that leg away. A conditional protective put comes on only when the book is levered past 1.10×. The sector-rotation sleeve (≤15%, unlevered, wide static stops) rebalances monthly. Zero single-name stock picking — that sleeve was retired after proving to be all cost and no fills.
Trade the open, survive always
A separate zero-LLM script trades opening-range breakouts on SPY — a small day-only sleeve that enters on a break of the first 30 minutes' range, trails a stop, and is always flat by the close. Meanwhile two intraday risk-checks (10:30 + 13:30 CT) plus midday enforce the survival stack: drawdown ladder (−10/−20/−25%), VIX kill-switch, margin-distance gate, fast-reversal trigger. A deadman watchdog flattens the book if the routines ever stall. Friday's Opus review runs per-sleeve alpha attribution and the kill-checks.
How it's built
No databases, no ORM, no in-memory state. Every memory file — strategy rulebook, trade ledger, daily research, the atomic risk-state file, weekly reviews — is committed to a private GitHub repo. Every routine run is a fresh container that clones, decides, commits, exits.
Hard rules the LLM cannot override
Leverage is set by formula and held inside a cage of survival mechanics enforced before any order is placed. The LLM runs the routines and picks the single names; the rules derive the core's leverage from realised volatility, cap it at 2×, keep the hedge on, and pull risk the moment drawdown or volatility crosses a line.
- Volatility-targeted leverage — gross = target ÷ realised vol, hard whole-book cap 2.0×
- Trend filter — leverage halved when the core is below its 200-day MA
- Hold the vol-target overnight — index returns live in the overnight leg; the old close-out habit measurably cost return
- Sector rotation sleeve — top 3 of 11 sector ETFs by 6-month momentum, monthly, ≤15% of equity, unlevered
- Intraday sleeve — opening-range breakouts on SPY, mechanical script, always flat by the close
- Conditional protective put — on only when the book is levered past 1.10× (premium ≤3% of NAV per year)
- Catastrophic GTC stop on the leveraged core; wide static stops on every rotation holding
- No single-name stock picking — the discretionary sleeve was retired on evidence, not feeling
- Drawdown ladder: −10% halve · −20% defensive · −25% hard halt; VIX kill-switch + 15% margin-distance gate
- Margin financing modelled net-of-carry; deadman watchdog flattens the book on any stall
What this actually is
This is a learning project. The infrastructure — an autonomous LLM agent, hard-rule risk discipline, git-as-memory, an atomic risk-state file — has engineering value regardless of P&L. There is no signal service, no subscription, nothing for sale, no community — just a private dashboard for the operator and a public page explaining what was built.
See it run
The dashboard is private. The only person with access is the operator. Click below if that's you.
Sign in