LLMs for Quant Research: Build Text Signals & Research Agents You Can Trust

Build financial research tools that extract trading signals, verify evidence, and evaluate investment strategies using LLMs, RAG, and reliable backtesting.

Workshop description

• Build practical research applications that extract financial signals from earnings calls, analyze SEC filings, and provide answers backed by verifiable sources.
• Develop research agents that use financial data and Python tools, then evaluate their accuracy, reliability, and potential investment value through structured testing.

###

Who this workshop is for

• **Data scientists, ML engineers, AI developers, and Python developers**looking to build financial research applications using LLMs, RAG, and machine learning.
• **Quantitative analysts, quant researchers, and algorithmic traders**interested in extracting trading signals, testing strategies, and improving research reliability.
• **Financial analysts, equity research analysts, and investment researchers** seeking to analyze earnings calls, company filings, and financial disclosures more efficiently.
• **Portfolio analysts, risk analysts, and investment professionals**interested in evaluating alternative data, validating financial signals, and supporting evidence-based investment decisions.

###

What you will learn

• **Financial text signals:** Extract sentiment, earnings guidance changes, and risk indicators from financial documents, benchmarking results against established financial language models.
• **RAG for financial research:** Build a filings research application that retrieves relevant information, interprets tables, and provides source-backed answers.
• **Research agents:** Connect LLMs to market data, financial filings, and Python tools with controlled access, spending limits, and human approval.
• **Reliable evaluation:** Use labelled datasets, incremental-information tests, and walk-forward backtesting to assess signal quality and prevent data leakage.

###

What you'll be able to build and evaluate

• A financial research pipeline that converts earnings calls into measurable signals and tests whether they add value beyond existing financial indicators.
• A research agent that analyzes SEC filings, retrieves verified information, and supports investment research without introducing future data into historical tests.

###

Prerequisites

• Working knowledge of Python, data analysis, and basic machine learning concepts.
• Familiarity with financial datasets, LLM APIs, or backtesting is helpful. A supported LLM API or local model is needed for hands-on exercises.

Key takeaways

• Build reusable research workflows that extract financial insights, validate evidence, and identify unreliable outputs.
• Evaluate trading signals with controlled backtests, prevent data leakage, and maintain reproducible research results.

What you will get

• Full HD workshop recording, certificate of participation, and presentation slides.
• Four working notebooks, approximately 200 labelled evaluation examples, and a pre-deployment checklist covering data leakage, evidence quality, incremental value, and reproducibility.

Meet your Instructor

• **Bruno Gonçalves**, Founder of Data For Science, author, trainer, and consultant specializing in Generative AI and Machine Learning.
• PhD in Physics of Complex Systems, former faculty member at Aix-Marseille Université, and former Data Science Fellow at NYU's Center for Data Science. He has taught thousands of practitioners through O'Reilly, Pearson, and live workshops.