Autonomous trading, human-approved.

A 3-state Gaussian HMM classifies the market regime each session and rotates between an equity sleeve (QQQ/SPY) and a defensive sleeve (Treasuries, gold, T-bills). Every decision — including a deliberate day in defense — is logged with its reasoning. Sign in to see what the model is doing today.

Regime rotation

Daily Viterbi inference on SPY log-returns, 21d realised vol, and VIX log-change drives rotation between risk (QQQ/SPY) and defense (TLT/GLD/BIL) by trailing momentum.

Slack-gated execution

No order routes to Alpaca without an explicit approval — even in autonomous mode, the audit trail is preserved.

Inactivity is data

A bear-regime day with no trade is still a logged decision with full reasoning. Survives the gap between trade and idle.