Paper: arXiv 2601.22119

Authors: Han Yang, Dong Hao, Zhuohan Wang, Qi Shi, Xingtong Li

Abstract

Automatically discovering formulaic alpha factors is a central problem in quantitative finance. Existing methods often ignore syntactic and semantic constraints, relying on exhaustive search over unstructured and unbounded spaces. We present AlphaCFG, a grammar-based framework for defining and discovering alpha factors that are syntactically valid, financially interpretable, and computationally efficient. AlphaCFG uses an alpha-oriented context-free grammar to define a tree-structured, size-controlled search space, and formulates alpha discovery as a tree-structured linguistic Markov decision process, which is then solved using a grammar-aware Monte Carlo Tree Search guided by syntax-sensitive value and policy networks. Experiments on Chinese and U.S. stock market datasets show that AlphaCFG outperforms state-of-the-art baselines in both search efficiency and trading profitability. Beyond trading strategies, AlphaCFG serves as a general framework for symbolic factor discovery and refinement across quantitative finance, including asset pricing and portfolio construction.

Complexity vs Empirical Score

  • Math Complexity: 6.5/10
  • Empirical Rigor: 8.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper provides a highly novel and mathematically elegant framework (AlphaCFG) that formalizes formulaic alpha generation using Context-Free Grammars and MCTS over a Tree-Structured Linguistic MDP. Empirical evaluation is rigorous, testing performance on major benchmarks (CSI 300, S&P 500) and demonstrating clear superiority over RL and GP baselines with provided open-source code.

Research Flowchart

  flowchart TD
    A["Research Goal: Discover valid, interpretable, and profitable alpha factors automatically"] --> B["Input Data: Chinese & U.S. Stock Market Datasets"]
    B --> C["AlphaCFG Framework: Define search space via Alpha-Oriented Context-Free Grammar"]
    C --> D["Formulate Discovery as Tree-Structured Linguistic MDP"]
    D --> E["Search & Optimization: Grammar-Aware MCTS with Syntax-Sensitive Policy/Value Networks"]
    E --> F["Outcomes: Superior Search Efficiency & Higher Trading Profitability"]

Cited References (44)

  1. The algebraic theory of context-free languages (1959)
  2. Report on the algorithmic language ALGOL 60 (1960)
  3. Autoalpha: an efficient hierarchical evolutionary algorithm for mining alpha factors in quantitative investment (2020) — link
  4. TreeNet: Learning Sentence Representations with Unconstrained Tree Structure. (2018)
  5. Mastering the game of go without human knowledge (2017)
  6. Monte-carlo tree search (2010)
  7. Improved semantic representations from tree-structured long short-term memory networks (2015) — link
  8. Active Portfolio Management: A Quantitative Approach for Providing Superior Returns and Controlling Risk (1999)
and 36 more
  1. XGBoost: A Scalable Tree Boosting System (2016)
  2. LightGBM: a highly efficient gradient boosting decision tree (2017)
  3. Detecting similar Java classes using tree algorithms (2006)
  4. 101 Formulaic Alphas (2016)
  5. Multi-Factor Stock Selection System based on Short-Period Price–Volume Features (Quantitative Topic No.93) (2017)
  6. On persistence in mutual fund performance (1997)
  7. The cross-section of expected stock returns (1992)
  8. Proximal policy optimization algorithms (2017) — link
  9. Interpretable scientific discovery with symbolic regression: a review (2024)
  10. Risk and return in an equilibrium APT: Application of a new test methodology (1988)
  11. Introduction to automata theory, languages, and computation (2001)
  12. Unsupervised grammar induction with depth-bounded PCFG (2018)
  13. Three Models for the Description of Language (1956)
  14. Automata Theory, Languages, and Computation (1979)
  15. Qlib: An ai-oriented quantitative investment platform (2020) — link
  16. Long short-term memory (1997)
  17. An empirical evaluation of generic convolutional and recurrent networks for sequence modeling (2018) — link
  18. Attention is all you need (2017)
  19. AlphaQCM: Alpha Discovery in Finance with Distributional Reinforcement Learning (2025)
  20. Generating synergistic formulaic alpha collections via reinforcement learning — → 2306.12964
  21. Active portfolio management (2000)
  22. Reverse Polish Notation Method (2004)
  23. Efficient selectivity and backup operators in Monte-Carlo tree search (2006)
  24. Stock price prediction using reinforcement learning (2001)
  25. Stock price prediction using ARIMA versus XGBoost models: the case of the largest telecommunication company in the Middle East (2023)
  26. Stock market prediction using linear regression and SVM (2021)
  27. Clustering approach to stock market prediction (2012)
  28. Finite-time analysis of the multiarmed bandit problem (2002)
  29. Actor-critic algorithms (1999)
  30. Assets Forecasting with Feature Engineering and Transformation Methods for LightGBM — → 2501.07580
  31. An XGBoost-based multivariate deep learning framework for stock index futures price forecasting (2023)
  32. Predicting stock market index using LSTM (2022)
  33. A dual-stage attention-based recurrent neural network for time series prediction (2017) — link
  34. Price change prediction of ultra high frequency financial data based on temporal convolutional network (2022)
  35. Predictive modeling of stock prices using transformer model (2024)
  36. Unsupervised temporal encoding for stock price prediction through dual-phase learning (2025)

2 of 44 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.