AiAgentForTrading

Most quantitative trading bots fail not because their signals are bad, but because market regimes shift and single-strategy models lack post-trade governance. When markets transition from low-volatility trends into sudden liquidity shocks, pure trend-following models suffer sharp drawdowns, while isolated contrarian bets bleed against persistent momentum. AiAgentForTrading approaches trading the way an institutional investment committee operates: through multi-perspective debate and deterministic risk gating. Built natively on the Alpaca Trading and Market Data APIs alongside LangGraph, the platform orchestrates five specialized strategy agents in parallel—Momentum (breakouts & volume), Macro Fundamentals (VIX & credit spreads), Statistical Arbitrage (SPY correlation & Z-score mean reversion), Contrarian (oversold extremes), and Derivatives (options flow & 24-hour cached earnings windows). A dynamic Regime Engine continuously assesses broader macro indicators to rebalance agent voting weights in real time. Only setups achieving a 70%+ weighted consensus advance to order generation. Execution is dual-track: depending on the setup and market regime, proposals are intelligently routed into either defined-risk Bull Call options spreads or Common Stock Bracket OCO orders with pre-calculated take-profit and stop-loss levels. Crucially, the system does not stop once an order is filled. A background Cron Sentinel runs continuous health audits: it automatically attaches missing protective OCO orders to orphan positions, market-closes option contracts before the theta decay cliff (DTE ≤ 2 days) to eliminate assignment risk, and sweeps idle cash into short-term Treasury ETFs ($SGOV) for risk-free baseline yield. Every decision passes through non-negotiable, code-level RiskGuard limits and writes directly to an immutable audit trail.