Vantage is an autonomous AI trading agent built on Alpaca's full developer stack — Trading API, MCP server, and CLI — designed for transparent, explainable algorithmic trading rather than black-box automation. At its core, Vantage runs a pairs-trading strategy on SPY/QQQ, continuously computing rolling z-scores on live market data pulled through Alpaca's MCP server. When a statistically significant spread divergence appears, the signal is handed to Google Gemini, which reasons over the market context and strategy state to decide whether to hold, enter, or exit a position — every decision logged in a full Reasoning Transparency Log showing the market signal, the AI's chain-of-thought, the MCP tools invoked, and the resulting action. Beyond the interactive MCP-driven agent loop, Vantage also uses the Alpaca CLI for scheduled automation tasks — end-of-day rebalancing and position monitoring — demonstrating all three pillars of Alpaca's stack working together: the Trading API for direct data access, MCP for the AI's interactive decision-making, and the CLI for lightweight scheduled jobs. The dashboard itself is built to feel like a real trading terminal: live candlestick charts with volume, a Binance-style scrolling price ticker, a positions and orders panel, and an analytics page reporting Sharpe ratio, win rate, and max drawdown — all computed honestly from real account history, explicitly showing "insufficient data" rather than fabricating numbers when trade history is too sparse. Every trade happens in Alpaca's paper trading environment — real market data, simulated funds, zero financial risk — making Vantage a safe, fully auditable demonstration of how an AI agent can trade with explainable reasoning at every step.
Category tags: