Theta Research rebuilds the institutional investment pipeline into an AI-native decision system for individual investors. Developed by a team fluent in Finance and ML. It researches the companies you own, maps their supply-chain dependencies, predicts and scores every earnings call, and tells you what changed and what to do. Numbers flow through deterministic code, not LLM guesses: EDGAR-reconciled financials, a 30,000-edge supply chain network, HF level algorithms and 72% hit rate.
Why pay for an AI research agent you could rebuild in an afternoon?
A few open-source skills, a few prompts, a polished interface. That is often the whole product: a chatbot that talks about stocks.
Theta is not a chatbot. It is a decision system. We rebuilt the full institutional investment pipeline, from information collection to research, prediction, and decision, into software that researches the companies you own, remembers why every position exists, and says what to do when things change. For $200 a year, individual investors get what used to require an $80,000 salary.
We spent our careers in finance and machine learning, and we built Theta around the three failures most equity agents never solve.
ā 1. Hallucination. LLMs only read, extract, and explain. Machine-learning models and a deterministic data pipeline do the real work: detect signals, compare historical samples, estimate event probabilities, and recalibrate against real outcomes. We know LLMs too well to trust them with numbers.
ā 2. Architecture. Theta follows the actual buy-side workflow end to end: idea discovery, thesis formation, financial modeling, management tracking, catalyst monitoring, estimate revisions, post-event review. Supply-chain intelligence is one source of edge inside that system: it maps real dependencies and traces how a change moves through related companies into your portfolio.
ā 3. Iteration. Every data point, assumption, and calculation is tracked. After each event, Theta scores its original estimate against the real outcome (68% hit rate across 370 scored earnings predictions so far), and the result calibrates the next call.
ā All of it converges at the decision layer. Every day, Theta answers four questions: What changed today? Is the thesis stronger or weaker? Which holdings are affected? What needs attention now?
ā It is live at theta.md. We will be in the comments all day, and we would especially love feedback from anyone who has tried to build one of these themselves.