Paper: arXiv 2403.00782

Abstract

Recent advancements in large language models (LLMs) have opened new pathways for many domains. However, the full potential of LLMs in financial investments remains largely untapped. There are two main challenges for typical deep learning-based methods for quantitative finance. First, they struggle to fuse textual and numerical information flexibly for stock movement prediction. Second, traditional methods lack clarity and interpretability, which impedes their application in scenarios where the justification for predictions is essential. To solve the above challenges, we propose Ploutos, a novel financial LLM framework that consists of PloutosGen and PloutosGPT. The PloutosGen contains multiple primary experts that can analyze different modal data, such as text and numbers, and provide quantitative strategies from different perspectives. Then PloutosGPT combines their insights and predictions and generates interpretable rationales. To generate accurate and faithful rationales, the training strategy of PloutosGPT leverage rearview-mirror prompting mechanism to guide GPT-4 to generate rationales, and a dynamic token weighting mechanism to finetune LLM by increasing key tokens weight. Extensive experiments show our framework outperforms the state-of-the-art methods on both prediction accuracy and interpretability.

Complexity vs Empirical Score

  • Math Complexity: 2.0/10
  • Empirical Rigor: 8.5/10
  • Quadrant: Street Traders — practical and empirical, lighter on theory

Why this score: The paper focuses on implementing and integrating existing deep learning and LLM techniques with novel prompting strategies for stock prediction, relying heavily on data processing, model fine-tuning, and backtesting results rather than presenting new theoretical mathematics.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Interpretable Stock Prediction"] --> B["Input Data<br>Financial Text & Numerical Data"]
  B --> C["PloutosGen<br>Multimodal Expert Fusion"]
  C --> D["PloutosGPT<br>LLM Rationale Generation"]
  D --> E["Training Mechanisms<br>Rearview-Mirror Prompting<br>Dynamic Token Weighting"]
  E --> F["Output<br>Accurate Prediction + Interpretable Rationale"]
  F --> G["Outcome<br>Outperforms SOTA<br>in Accuracy & Interpretability"]

Cited References (146)

  1. Structure and Interpretation of Computer Programs (1985)
  2. Visual Information Extraction with Lixto (2001)
  3. An overview of the KL-ONE knowledge representation system (1985)
  4. Complexity results for nonmonotonic logics (1992)
  5. Hypertree Decompositions and Tractable Queries (2002)
  6. Foundations of a functional approach to knowledge representation (1984)
  7. A logic of implicit and explicit belief (1984)
  8. On the compilability and expressive power of propositional planning formalisms (2000)
and 138 more
  1. IJCAI Camera Ready Submission
  2. FinGPT: Open-Source Financial Large Language Models — → 2306.06031
  3. Hybrid deep sequential modeling for social text-driven stock prediction (2018)
  4. Halueval: A large-scale hallucination evaluation benchmark for large language models (2023)
  5. Truthfulqa: Measuring how models mimic human falsehoods (2021) — link
  6. Causality-Guided Multi-Memory Interaction Network for Multivariate Stock Price Movement Prediction (2023)
  7. Enhancing stock movement prediction with adversarial training (2018) — link
  8. Explainable text-driven neural network for stock prediction (2018)
  9. A Novel ensemble neuro-fuzzy model for financial time series forecasting (2019)
  10. An interpretable neuro-fuzzy approach to stock price forecasting (2019)
  11. Applications of Explainable Artificial Intelligence in Finance—a systematic review of Finance, Information Systems, and Computer Science literature (2023)
  12. Large Language Models in Finance: A Survey (2023)
  13. Bloomberggpt: A large language model for finance (2023) — link
  14. Instruct-FinGPT: Financial Sentiment Analysis by Instruction Tuning of General-Purpose Large Language Models — → 2306.12659
  15. Promptcast: A new prompt-based learning paradigm for time series forecasting (2023)
  16. Meta-transformer: A unified framework for multimodal learning (2023) — link
  17. One Fits All: Power General Time Series Analysis by Pretrained LM (2023) — link
  18. ChatGPT Informed Graph Neural Network for Stock Movement Prediction — → 2306.03763
  19. Alpha-GPT: Human-AI Interactive Alpha Mining for Quantitative Investment — → 2308.00016
  20. Gpt-4 technical report (2023) — link
  21. Llama 2: Open foundation and fine-tuned chat models (2023) — link
  22. Listening to chaotic whispers: A deep learning framework for news-oriented stock trend prediction (2018)
  23. Topic modeling based sentiment analysis on social media for stock market prediction (2015)
  24. Stock price prediction using LSTM, RNN and CNN-sliding window model (2017)
  25. Smoothing, forecasting and prediction of discrete time series (2004)
  26. The Wall Street Neophyte: A Zero-Shot Analysis of ChatGPT Over MultiModal Stock Movement Prediction Challenges — → 2304.05351
  27. Goat: Fine-tuned LLaMA Outperforms GPT-4 on Arithmetic Tasks (2023) — link
  28. Rest: Relational event-driven stock trend forecasting (2021)
  29. Twitter Financial News Sentiment (2022)
  30. Bitcoin Tweets Sentiment Analysis (Accessed: 2023)
  31. Data-Centric Financial Large Language Models (2023) — link
  32. Www'18 open challenge: financial opinion mining and question answering (2018)
  33. Exploiting social relations and sentiment for stock prediction (2014)
  34. Exploiting topic based twitter sentiment for stock prediction (2013)
  35. FAST: Financial news and tweet based time aware network for stock trading (2021)
  36. Stock movement prediction from tweets and historical prices (2018)
  37. Hats: A hierarchical graph attention network for stock movement prediction (2019) — link
  38. Spatiotemporal hypergraph convolution network for stock movement forecasting (2020)
  39. StockFormer: learning hybrid trading machines with predictive coding (2023)
  40. A CNN-BiLSTM-AM method for stock price prediction (2021)
  41. CNNpred: CNN-based stock market prediction using a diverse set of variables (2019)
  42. Stock market’s price movement prediction with LSTM neural networks (2017)
  43. A dual-stage attention-based recurrent neural network for time series prediction (2017) — link
  44. Stock price prediction using attention-based multi-input LSTM (2018)
  45. Large language models are zero-shot time series forecasters (2023) — link
  46. Letter to Shareholders (1991). (1991)
  47. A survey on large language model based autonomous agents (2023) — link
  48. Autogen: Enabling next-gen llm applications via multi-agent conversation framework (2023) — link
  49. PIXIU: A Large Language Model, Instruction Data and Evaluation Benchmark for Finance (2023) — link
  50. Can chatgpt forecast stock price movements? return predictability and large language models — → 2304.07619
  51. Deep attentive learning for stock movement prediction from social media text and company correlations (2020)
  52. An empirical analysis of data requirements for financial forecasting with neural networks (2001)
  53. Do measures of investor sentiment predict returns? (1998)
  54. Stock price prediction using prior knowledge and neural networks (1997)
  55. Association between accounting performance measures and stock prices: A test of the life cycle hypothesis (1992)
  56. Stock market volatility and learning (2016)
  57. Integrating knowledge tracing and item response theory: A tale of two frameworks (2014)
  58. Knowledge tracing: Modeling the acquisition of procedural knowledge (1994)
  59. Question Difficulty Prediction for READING Problems in Standard Tests (2017)
  60. EKT: Exercise-aware Knowledge Tracing for Student Performance Prediction (2019)
  61. Deep knowledge tracing (2015)
  62. Dynamic key-value memory networks for knowledge tracing (2017)
  63. Graph-based Knowledge Tracing: Modeling Student Proficiency Using Graph Neural Network (2019)
  64. Bert: Pre-training of deep bidirectional transformers for language understanding (2018) — link
  65. Convolutional neural networks for sentence classification (2014) — link
  66. What Does BERT Look At? An Analysis of BERT’s Attention (2019) — link
  67. Long short-term memory (1997)
  68. Interpretable Cognitive Diagnosis with Neural Network for Intelligent Educational Systems (2019) — link
  69. Hierarchical clustering schemes (1967)
  70. The choice of item difficulty in self-adapted testing. (2000)
  71. Xgboost: A scalable tree boosting system (2016)
  72. node2vec: Scalable feature learning for networks (2016)
  73. What Does a TextCNN Learn? (2018) — link
  74. Root mean square error (RMSE) or mean absolute error (MAE)?–Arguments against avoiding RMSE in the literature (2014)
  75. The effect of analogy instruction on young children’s metaphor comprehension (1998)
  76. The relative effectiveness of human tutoring, intelligent tutoring systems, and other tutoring systems (2011)
  77. Efficient graph representations (2003)
  78. Using AUC and accuracy in evaluating learning algorithms (2005)
  79. Addressing two problems in deep knowledge tracing via prediction-consistent regularization (2018)
  80. Exploring hierarchical structures for recommender systems (2018)
  81. Hierarchical lifelong learning by sharing representations and integrating hypothesis (2019)
  82. Performance Factors Analysis–A New Alternative to Knowledge Tracing. (2009)
  83. Engaging with massive online courses (2014)
  84. Ubiquitous learning environment: An adaptive teaching system using ubiquitous technology (2004)
  85. Intelligent tutoring systems: Evolutions in design (2014)
  86. Hierarchical graph representation learning with differentiable pooling (2018)
  87. Incorporating rich features into deep knowledge tracing (2017)
  88. Student diagnosis in practice; bridging a gap (1995)
  89. Finding similar exercises in online education systems (2018)
  90. Retaining Minority Students in Higher Education: A Framework for Success. ASHE-ERIC Higher Education Report. Jossey-Bass Higher and Adult Education Series. (2003)
  91. Exam question recommender system (2005)
  92. Similarity Measure of Test Questions Based on Ontology and VSM (2014)
  93. A reinforcement learning approach to personalized learning recommendation systems (2019)
  94. Recommendation system for adaptive learning (2018)
  95. Enhancing mathematical problem solving among third-grade students with schema-based instruction. (2004)
  96. word2vec Explained: deriving Mikolov et al.’s negative-sampling word-embedding method (2014) — link
  97. Deep contextualized word representations (2018) — link
  98. Disan: Directional self-attention network for rnn/cnn-free language understanding (2018)
  99. MAM-RNN: Multi-level Attention Model Based RNN for Video Captioning. (2017)
  100. Adam: A method for stochastic optimization (2015)
  101. The graph neural network model (2008)
  102. Graph neural networks: A review of methods and applications (2018) — link
  103. Interoperability in personalized adaptive learning (2006)
  104. Textrank: Bringing order into text (2004)
  105. Silhouettes: a graphical aid to the interpretation and validation of cluster analysis (1987)
  106. Some methods for classification and analysis of multivariate observations (1967)
  107. A density-based algorithm for discovering clusters in large spatial databases with noise. (1996)
  108. A survey of text clustering algorithms (2012)
  109. Data clustering and similarity (2013)
  110. A comprehensive survey on graph neural networks (2020)
  111. Semi-supervised classification with graph convolutional networks (2017) — link
  112. Personalized recommendation of learning material using sequential pattern mining and attribute based collaborative filtering (2014)
  113. Exploiting cognitive structure for adaptive learning (2019)
  114. Automatically learning topics and difficulty levels of problems in online judge systems (2018)
  115. Cognitive modelling for predicting examinee performance (2015)
  116. Cognitive diagnostic assessment for education: Theory and applications (2007)
  117. DINA model and parameter estimation: A didactic (2009)
  118. Probabilistic-input, noisy conjunctive models for cognitive diagnosis (2018)
  119. The gap of semantic parsing: A survey on automatic math word problem solvers (2019)
  120. Natural language input for a computer problem solving system (1964)
  121. Experiments with a deductive question-answering program (1965)
  122. Understanding and solving arithmetic word problems: A computer simulation (1985)
  123. Robust understanding of word problems with extraneous information (2007)
  124. Frame-based calculus of solving arithmetic multi-step addition and subtraction word problems (2010)
  125. Neural message passing for quantum chemistry (2017) — link
  126. Inductive representation learning on large graphs (2017)
  127. Graph attention networks (2017) — link
  128. Knowledge questions from knowledge graphs (2017)
  129. Transition-based knowledge graph embedding with relational mapping properties (2014)
  130. Long-tail relation extraction via knowledge graph embeddings and graph convolution networks (2019) — link
  131. Case studies in the use of ROC curve analysis for sensor-based estimates in human computer interaction (2005)
  132. Stable prediction across unknown environments (2018)
  133. A comparison of document clustering techniques (2000)
  134. Empirical and theoretical comparisons of selected criterion functions for document clustering (2004)
  135. Evaluation of hierarchical clustering algorithms for document datasets (2002)
  136. Fast and Intuitive Clustering of Web Documents. (1997)
  137. Memory: A contribution to experimental psychology (2013)
  138. Self-attention graph pooling (2019) — link

6 of 146 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.