QuantCoach: An Open-Source Alpha Research Copilot for My Quant Friend

A developer built QuantCoach, an open-source, voice-enabled AI research copilot for a friend working through WorldQuant BRAIN's alpha research platform. The tool chains ElevenLabs speech-to-text and text-to-speech with Groq-hosted open-weight models (GPT-OSS 120B) and Backboard persistent memory to turn spoken alpha intuitions into BRAIN-style Fast Expressions with explanations, while keeping proprietary factor ideas off closed APIs. The developer argues open-weight, portable models are essential for quant researchers who cannot risk their hypotheses entering a vendor's training pipeline.

This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend https://dev.to/challenges/hacktoberfest-weekend-2026-10-01 What I Built I built QuantCoach , a voice-enabled open-source AI research copilot for my friend who is grinding through WorldQuant BRAIN. My friend was stuck in a specific loop: they'd sketch an alpha idea on paper, try to write it as a BRAIN Fast Expression, get stuck on the math or the operator syntax, and lose momentum. The next day they'd forget why the previous attempt failed. Their research notebook was a graveyard of half-finished ideas with no thread connecting them. QuantCoach fixes this. It's a voice-first workspace where they: - Describe an alpha intuition out loud "combine improving margins with a short-term reversal filter" . - Get back a candidate BRAIN-style Fast Expression plus the math behind it. - Hear the explanation read aloud so they can keep their hands on the keyboard. - Build a permanent memory of which factor families worked and which ones burned them. - Ask follow-up DSA/math questions in the same session without breaking flow. It's a thinking partner for alpha ideation and quantitative learning, not a black box that promises returns. Demo 🚀 Try the Live Demo on Render https://quantcoach.onrender.com/ Note: Hosted on Render's free tier, so it may take 30–60 seconds to wake up on the first visit. Thanks for your patience Code 🔗 GitHub Repository https://github.com/aaryangif/QuantCoach How I Built It QuantCoach is built with open-source AI at its core , using a modular stack designed for privacy and speed: - Groq — Ultra-low-latency inference on open-weight models GPT-OSS 120B . Fast enough that the voice loop feels conversational. - Backboard — Persistent Memory & RAG. Every alpha idea, every DSA question, every "that one didn't work" gets stored so the coach remembers the researcher's history. - ElevenLabs — Scribe STT for voice input, multilingual TTS for reading back explanations and BRAIN expressions. This is what makes hands-free research actually usable. - MongoDB Atlas Optional — Powers the weak-spot tracker and progress dashboard. If not configured, the app falls back to in-memory storage with no crashes. - Streamlit — The entire interface, styled with a warm almond/crimson "editorial quant" aesthetic and a custom Lucide SVG icon system. The alpha generation flow is a straight pipeline: Voice → ElevenLabs STT → Backboard Memory → Groq LLM BRAIN-style prompt → Fast Expression + reasoning → Backboard save → ElevenLabs TTS → Voice Out. The alpha generator uses a specialized system prompt that: - Translates natural language intuition into BRAIN Fast Expression syntax - Calls out common pitfalls lookahead leakage, turnover drag, overfitting across lookback windows - Explicitly warns when an idea is likely to be correlated with known industry factors - Refuses to promise performance — it's an ideation tool, not a signal service Why Does Open Innovation Matter? Open-source AI wasn't a checkbox here. It was the entire reason this project is safe for a quant researcher to use. 1. Alpha ideas are proprietary. This is the big one. My friend is developing factor hypotheses that could be worth real money. Sending those prompts to a closed API means they enter someone else's training pipeline. Running open-weight models through Groq and storing memory in Backboard means the research context stays private. This is non-negotiable for anyone doing serious quant work. 2. Model portability is a research edge. The alpha landscape moves fast. When a better open-weight model drops, they swap GROQ MODEL NAME in .env and the whole system upgrades. No rewrite. No negotiation with a closed vendor's roadmap. 3. Zero cost to the learner. Quant education is already gated by expensive data subscriptions and paid courses. The tooling shouldn't add to that. The entire QuantCoach stack runs on free tiers, making it accessible to anyone with curiosity and a laptop. 4. Transparency matters for a coach. Because the model is open-weight and the prompts are in the repo utils/prompts.py , anyone can inspect exactly how the coach reasons about alphas. With a closed API, that reasoning is a black box. Prize Categories - Best Use of Render — Frontend and AI runtime hosted on Render - Best Use of Backboard — Memory & RAG powering the persistent research context - Best Use of ElevenLabs — Full voice loop STT + TTS for hands-free alpha research - Best Use of MongoDB Atlas — Optional persistence layer for the weak-spot tracker - Best Use of Groq — Open-weight model inference for real-time alpha ideation TY for this ossum opportunity...