SentinelAlpha

SentinelAlpha is an explainable AI portfolio-protection agent that detects elevated investment risk and deploys protective put options through Alpaca Paper Trading. Instead of simply recommending that investors sell concentrated positions during periods of uncertainty, SentinelAlpha determines whether an options hedge is appropriate, how much protection is needed, and which contract best fits the portfolio’s risk profile. Four specialized agents analyze market trends, quantitative indicators, financial news, and portfolio exposure. Their findings are combined into a transparent risk assessment covering concentration, volatility, negative momentum, news sentiment, drawdown, and defensive agent consensus. When risk exceeds the configured threshold, the Hedge Agent selects a tradable put based on expiration, strike distance, liquidity, premium cost, and available options buying power. The system then calculates a bounded hedge ratio and sizes the order against actual 100-share position lots. Before execution, every proposal must pass 17 deterministic risk controls. These verify paper mode, portfolio eligibility, activation thresholds, contract validity, expiration, strike distance, hedge size, premium limits, liquidity, quote freshness, duplicate orders, trading frequency, market hours, semantic risk review, kill-switch status, Alpaca execution interface, and account permissions. AI agents can provide reasoning, but they cannot bypass these controls. Approved hedges can be executed autonomously through Alpaca’s official CLI using BUY_TO_OPEN limit orders. SentinelAlpha is permanently restricted to paper trading and supports supervised operation when human approval is preferred. Every decision produces an ExplainHedge record containing the agents’ evidence, dissenting opinions, risk score, selected contract, rejected or approved controls, execution result, and pre-committed release and rebalancing conditions.