Maximus AI

Maximus is an autonomous paper trading system built for the Alpaca AI Trading Agents Hackathon. Its goal is to identify short-term opportunities in technology, SaaS, semiconductors, infrastructure, robotics and energy while keeping every decision traceable. The system continuously collects Alpaca market data, option chains, company news, sector news, official company feeds, SEC filings, Google News RSS and GDELT. It continues monitoring outside market hours so that scenarios are prepared before the next session. Each analysis packet includes recent price action, volume, moving averages, sector performance, historical context, upcoming events, liquidity and current portfolio exposure. Specialized agents review news, market structure and sector behavior. A strategist proposes one action, a critic searches for contradictory evidence, and a principal agent makes the final decision. The available actions are Long, Short, Call, Put, Close and Hold. The language model never sends an order directly. Every proposal must pass a deterministic risk engine. It checks account freshness, quote freshness, spreads, cash reserves, position limits, maximum loss, market hours and exit conditions. Only the isolated execution gateway can submit an approved paper order through Alpaca. Rejected proposals and their reasons remain in the audit trail. The project uses Alpaca Trading, Market Data, News and Options APIs, together with a read-only Alpaca MCP capability audit and isolated model advisors. A secured live dashboard presents P&L, positions, news sources, decisions, risk checks, data freshness and AI usage. The paper account started at $100,000 and showed $100,434.37 in equity at the submission snapshot.