🤝 Top Collaborators
.png&w=640&q=75)
🤓 Latest Submissions
Should-AI Buy?
THE PROBLEM WITH AI TRADING BOTS Most AI trading bots rely on a single prompt asking an LLM to pick assets. During market volatility, they hallucinate, chase false breakouts, and liquidate accounts without risk discipline. OUR SOLUTION: THE ADVERSARIAL AI COUNCIL Should-AI Buy? is an autonomous multi-agent system asking: "Why should we NOT take this trade?" Instead of a single bot rushing to buy, our system runs a council of specialized agents: • Momentum & Trend Agents: Detect market regime shifts, volume spikes, and breakout candidates. • Adversarial Red Team: A dedicated skeptic agent that stress-tests theses and kills bad trades before risking capital. • Deterministic Risk Gate: LLMs generate theses, but never place direct orders. Every trade must clear strict quantitative filters: ATR invalidation levels, a 75% max portfolio exposure cap, and position limits. PRODUCTION-GRADE ARCHITECTURE & 24/7 AUTONOMY • Live Broker Ground Truth: Integrated with Alpaca Paper Trading API v2 with sub-40ms reconciliation. • 24/7 AWS Cloud Daemon: Running continuously under PM2 clusters on AWS EC2. • Dual-Role Security Gate: Role separation with 1-Click Judge Access (Passphrase: alpaca2026) for safe read-only inspection. • Dynamic Risk Slider: A continuous 0%–100% human-in-the-loop capital dial to adjust risk tolerance in real time. • Interactive Equity Curve: Custom TradingView-style portfolio chart with crosshair tooltips and P&L tracking. • Engineering Rigor: Backed by 891 automated unit tests with a 100% pass rate. Live on AWS: http://15.134.249.209:3000 (Click "1-Click Judge Access" to unlock instant View-Only evaluation).
4 Sep 2026
