Alpha Judge: Prove the Trade Before You Trade It

The Problem: Most AI trading bots ask one flawed question: "What should we buy?" They blindly execute trades based on LLM hallucinations, often leading to blown accounts due to poor liquidity, wide spreads, or out-of-hours trading. The Solution: Alpha Judge Alpha Judge asks a different question: "Can we mathematically prove this trade deserves our capital?" It is a strict, hedge-fund-grade quantitative terminal that treats AI as a research analyst, not a reckless executor. How it Works: Multi-Perspective Analysis: We fetch live market data via the Alpaca API and feed it to LLMs (via OpenRouter) to generate a trading signal and confidence score. The Trade Proof Engine (Risk Gates): This is the core innovation. Even if the AI confidently screams "BUY", our deterministic backend evaluates the live Alpaca options chain. If the Bid/Ask spread is too wide, or no contracts match our DTE safety window, the trade is ruthlessly BLOCKED. It protects capital by knowing when not to trade. Context-Aware Agent ("Ask Alpha Judge"): To ensure absolute transparency, we built a floating real-time chatbot. Users can ask the agent why a specific trade was blocked, and it reads the live JSON state of the dashboard to explain exactly which risk parameter failed. Tech Stack: Built with a Python/Flask risk-engine backend, a Next.js/Tailwind dark-mode terminal UI, Alpaca Trading/Data APIs, and OpenRouter for LLM reasoning.