leoncuhk/awesome-quant-ai

what: A curated list (an "awesome" style list) collecting resources for quantitative investing that uses AI and machine learning. It links to papers, open-source tools, books, and courses, rather than being a tool itself. The README also includes original explanatory content about how to design trading strategies, compare different trading approaches, and what's new in the field for 2025/2026. An associated site, “Regime Research,” publishes daily data on market regimes, volatility, and risk across 30 tickers.

core ideas the list explains:

• Three pillars of quant finance: The main challenges (e.g., market efficiency, whether factors like “momentum” persist, trading costs), which AI techniques map to each problem (predicting returns, finding hidden patterns, learning trading policies from interaction, generating synthetic market data, understanding news and filings), and the math foundation (stochastic processes, optimization, game theory).

• A design process for building a trading system: The README lays out seven steps — define objectives and constraints; research strategy ideas (from theory or data mining); build and calibrate models (watching for overfitting); backtest rigorously with realistic costs; integrate risk management; deploy with proper infrastructure; and continuously monitor and iterate.

• A taxonomy of trading strategies: Ten categories, each with sub-strategies and recommended source papers or tools — e.g., statistical arbitrage (mean reversion, pairs trading), factor investing (value, momentum, size), high-frequency trading (market making, latency arbitrage), trend following, volatility trading, risk parity, macro strategies, event-driven (merger arbitrage, earnings), machine-learning strategies (citing Stefan Jansen's 150+ notebook repository), and multi-strategy approaches (including Hierarchical Risk Parity).

• A comparison of three trading paradigms: A table contrasting Quantitative Trading (static rules, high transparency), Algorithmic Trading (predefined logic with parameter optimization, medium adaptability), and AI-Agent Trading (autonomous learning agents, high adaptability but “black box” decisions).

• Frontier topics (2025/2026): Two emerging areas. First, LLM-based trading agents — multi-agent frameworks where large language models play roles like analyst, trader, and risk manager (the list includes TradingAgents, FinRobot, FinGPT, FinRL, Vibe-Trading, oracle3). Second, transformer time-series foundation models — pre-trained models that forecast prices or volatility without task-specific training (Chronos, TimesFM, Moirai, Lag-Llama, PatchTST, TimeGPT, Kronos; the README text cuts off mid-entry).

other sections: The list also includes learning resources, books, research papers, original notes by the author, community and conference links, and related lists. The README opens with an image titled “Investment Research: The Complete Map" and poses the question “Your edge: which layer do you understand better than consensus?”

tone: Aimed at practitioners — the writing assumes comfort with finance and ML concepts — but the list's value is in pointing readers to primary sources (SSRN papers, GitHub repos, data libraries like the Kenneth French Data Library) so a learner can go deeper on any single item.