.png&w=256&q=75)
🤓 Latest Submissions
VolGuard AI
VolGuard AI is an autonomous options desk built on Alpaca paper trading. It scans SPY, QQQ, IWM, and GLD; compares realized and implied movement; and records both candidates and abstentions. Exact defined-risk legs are priced conservatively before a five-role specialist council and a fourteen-gate deterministic governor decide whether the proposal may proceed. Its D1-backed Decision Memory specialist requires repeated agreement across recent open-market scans, so a first sighting or conflicting analysis cannot trigger a new paper entry. A deterministic payoff-quality gate rejects capped structures offering less than $0.25 theoretical maximum profit for each $1.00 of maximum loss. Approved entries are validated through the Alpaca CLI as atomic multi-leg dry runs before an explicitly unlocked paper submission. A Cloudflare-dispatched cycle monitors positions every five minutes and scans for a new fully governed entry every ten minutes. Broker legs are reconciled to an append-only strategy ledger, and the monitor independently applies profit, loss, and pre-expiration time exits to as many as two open strategies. Live routing is prohibited. The deployed dashboard makes the full reasoning chain inspectable across Decisions, Positions, Risk, and Journal views. Sanitized telemetry is stored in Cloudflare D1, while brokerage credentials remain confined to the paper-only runner. VolGuard demonstrates that a useful trading agent is not merely able to place orders; it can explain its evidence, cap its loss before entry, refuse poor setups, and preserve a reviewable history for deliberate practice
4 Sep 2026
