This project developed a long/short equity strategy combining traditional quantitative methods with LLM integration. The team built and evaluated three core strategies — options-implied sentiment signals, mean reversion and pairs trading, and a commodities risk premium model — using U.S. equities across 1–5 week hold periods. The architecture centers on a risk matrix and alpha matrix framework, where an LLM pipeline processes daily news, SEC filings, earnings call transcripts, and macro indicators across the full Russell 3000 universe. The LLM outputs a scalar signal per ticker that feeds directly into a portfolio optimizer to produce a long/short basket. Paper trading results over a four-month sample demonstrated the framework as a proof of concept, with identified next steps including expanding the options research layer and building a more proprietary LLM solution.