Paper: arXiv 2311.03283

Authors: Haoyang Cao, Haotian Gu, Xin Guo, Mathieu Rosenbaum

Abstract

Transfer learning is an emerging and popular paradigm for utilizing existing knowledge from previous learning tasks to improve the performance of new ones. In this paper, we propose a novel concept of transfer risk and and analyze its properties to evaluate transferability of transfer learning. We apply transfer learning techniques and this concept of transfer risk to stock return prediction and portfolio optimization problems. Numerical results demonstrate a strong correlation between transfer risk and overall transfer learning performance, where transfer risk provides a computationally efficient way to identify appropriate source tasks in transfer learning, including cross-continent, cross-sector, and cross-frequency transfer for portfolio optimization.

Complexity vs Empirical Score

  • Math Complexity: 6.5/10
  • Empirical Rigor: 6.0/10 (capped from 7.0: transaction costs not modelled; see penalty flags)
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper introduces a formal mathematical framework for ’transfer risk’ in quantitative finance and validates it empirically across multiple asset domains. Its application of transfer learning to cross-sector and cross-continent portfolio optimization is both novel and rigorously tested.

Research Flowchart

  flowchart TD
    A[Research Goal: Evaluate Transferability & Risk in Financial Transfer Learning] --> B[Data Inputs: Stock Returns & Portfolios<br> Cross-continent, cross-sector, cross-frequency]
    B --> C[Methodology: Propose Transfer Risk Concept & Framework]
    C --> D[Computational Process: Model Stock Prediction & Portfolio Optimization]
    D --> E[Analysis: Evaluate Correlation Between Transfer Risk & Performance]
    E --> F[Outcome: Transfer Risk Efficiency Identifies Best Source Tasks for Target Transfer]

Cited References (119)

  1. Bandits for Algorithmic Trading with Signals (2023)
  2. Dynamically consistent alpha-maxmin expected utility (2020)
  3. Arbitrage-free modeling under Knightian uncertainty (2020)
  4. Uncertainty and filtering of hidden Markov models in discrete time (2020)
  5. Uncertain portfolio optimization problem under a minimax risk measure (2019)
  6. Uncertain minimax mean-variance and mean-semivariance models for portfolio selection 1 (2022)
  7. Duality theory for robust utility maximisation (2021)
  8. Utility maximization with proportional transaction costs under model uncertainty (2020)
and 111 more
  1. Statistical decision theory and Bayesian analysis (2013)
  2. Minimax Point Estimation (1950)
  3. On the minimax estimator of a bounded normal mean (2002)
  4. The Problem of Estimation (1957) — link
  5. Theory of point estimation (2006)
  6. Feasibility and Transferability of Transfer Learning: A Mathematical Framework (2023) — link
  7. Deep signature transforms (2019)
  8. The signature of a rough path: uniqueness (2016)
  9. Regularized Learning for Domain Adaptation under Label Shifts (2019)
  10. Transferability and hardness of supervised classification tasks (2019)
  11. L og ME : Practical assessment of pre-trained models for transfer learning (2021)
  12. Frustratingly easy transferability estimation (2022)
  13. Adversarial feature augmentation for unsupervised domain adaptation (2018)
  14. Multi-level semantic feature augmentation for one-shot learning (2019)
  15. A simple feature augmentation for domain generalization (2021)
  16. Transferability estimation using B hattacharyya class separability (2022)
  17. Minimax lower bounds for transfer learning with linear and one-hidden layer neural networks (2020)
  18. LEEP : A New Measure to Evaluate Transferability of Learned Representations (2020)
  19. Generalization Bounds for Deep Transfer Learning Using Majority Predictor Accuracy (2022)
  20. Generalization Bounds for Transfer Learning with Pretrained Classifiers (2022) — link
  21. Training language models to follow instructions with human feedback (2022)
  22. GPT-4 Technical Report (2023) — link
  23. Jointly modeling transfer learning of industrial chain information and deep learning for stock prediction (2022) — link
  24. FinBERT: A Pre-Trained Financial Language Representation Model for Financial Text Mining (2021)
  25. Natural language based financial forecasting: a survey (2018)
  26. Crude oil price prediction model with long short term memory deep learning based on prior knowledge data transfer (2019)
  27. Universal features of price formation in financial markets: perspectives from deep learning (2019) — link
  28. A novel approach to short-term stock price movement prediction using transfer learning (2019)
  29. A Survey on Transfer Learning (2010)
  30. A survey on deep transfer learning (2018)
  31. A comprehensive survey on transfer learning (2020)
  32. Sinkhorn distances: Lightspeed computation of optimal transport (2013)
  33. Transportation cost for Gaussian and other product measures (1996)
  34. Learning bounds for domain adaptation (2007)
  35. A theory of learning from different domains (2010)
  36. Tightening mutual information-based bounds on generalization error (2020)
  37. On learning invariant representations for domain adaptation (2019)
  38. On pairs of $ f $-divergences and their joint range (2011)
  39. A Mathematical Framework for Quantifying Transferability in Multi-source Transfer Learning (2021)
  40. Transfer Learning for Retinal Vascular Disease Detection: A Pilot Study with Diabetic Retinopathy and Retinopathy of Prematurity (2022) — link
  41. Learning a functional control for high-frequency finance (2020) — link
  42. Deep calibration of the quadratic rough Heston model (2021) — link
  43. Transfer learning in natural language processing (2019)
  44. Learning from the past, predicting the statistics for the future, learning an evolving system (2013) — link
  45. Conditional sig-wasserstein gans for time series generation (2020) — link
  46. Integration of paths–A faithful representation of paths by noncommutative formal power series (1958)
  47. Differential equations driven by rough paths (2007)
  48. Uniqueness for the signature of a path of bounded variation and the reduced path group (2010)
  49. Characteristic functions of measures on geometric rough paths (2016)
  50. A path signature approach for speech emotion recognition (2019)
  51. Transfer learning from speaker verification to multispeaker text-to-speech synthesis (2018)
  52. Analysis of causative attacks against svms learning from data streams (2017)
  53. On the optimality of conditional expectation as a Bregman predictor (2005)
  54. Impossibility theorems for domain adaptation (2010)
  55. An information-theoretic approach to transferability in task transfer learning (2019)
  56. Adapting visual category models to new domains (2010)
  57. Unsupervised domain adaptation by backpropagation (2015)
  58. Domain-adversarial training of neural networks (2016)
  59. Deep visual domain adaptation: A survey (2018)
  60. A survey of sentiment analysis based on transfer learning (2019)
  61. Instance Weighting for Domain Adaptation in NLP (2007)
  62. Sparse autoencoder-based feature transfer learning for speech emotion recognition (2013)
  63. Automatic ICD-9 coding via deep transfer learning (2019)
  64. Transfer learning for medical image classification: A literature review (2022)
  65. A heterogeneous label propagation algorithm for disease gene discovery (2010)
  66. Transfer learning in collaborative filtering for sparsity reduction (2010)
  67. DARec: deep domain adaptation for cross-domain recommendation via transferring rating patterns (2019)
  68. Deep-learning domain adaptation techniques for credit cards fraud detection (2020)
  69. Universal Language Model Fine-tuning for Text Classification (2018)
  70. Vl-adapter: Parameter-efficient transfer learning for vision-and-language tasks (2022)
  71. Transfer learning for activity recognition: A survey (2013)
  72. Stratified transfer learning for cross-domain activity recognition (2018)
  73. Adversarial discriminative domain adaptation (2017)
  74. Image N et: A large-scale hierarchical image database (2009)
  75. Transfer joint matching for unsupervised domain adaptation (2014)
  76. Optimal transport for domain adaptation (2017)
  77. O TCE : A Transferability Metric for Cross-Domain Cross-Task Representations (2021)
  78. Decision support from financial disclosures with deep neural networks and transfer learning (2017)
  79. Improving financial trading decisions using deep Q-learning: Predicting the number of shares, action strategies, and transfer learning (2019)
  80. Structured Pruning Learns Compact and Accurate Models (2022)
  81. An equity fund recommendation system by combing transfer learning and the utility function of the prospect theory (2018)
  82. Ammus: A survey of transformer-based pretrained models in natural language processing (2021) — link
  83. Cross-domain MLP and CNN transfer learning for biological signal processing: EEG and EMG (2020)
  84. A review on transfer learning in EEG signal analysis (2021)
  85. DeCAF: A Deep Convolutional Activation Feature for Generic Visual Recognition (2014)
  86. Learning transferable features with deep adaptation networks (2015)
  87. Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs (2016)
  88. Meta dynamic pricing: Transfer learning across experiments (2022)
  89. On the theory of transfer learning: The importance of task diversity (2020)
  90. Transferability Estimation for Semantic Segmentation Task (2021) — link
  91. Meta-learning with GANs for anomaly detection, with deployment in high-speed rail inspection system (2022) — link
  92. BERT : Pre-training of Deep Bidirectional Transformers for Language Understanding (2019)
  93. Least squares after model selection in high-dimensional sparse models (2013)
  94. Optimal two-step prediction in regression (2017)
  95. Regression shrinkage and selection via the lasso (1996)
  96. On model selection consistency of Lasso (2006)
  97. The adaptive lasso and its oracle properties (2006)
  98. Iterated Integrals and Exponential Homomorphism (1954)
  99. Integration of Paths, Geometric Invariants and a Generalized Baker-Hausdorff Formula (1957)
  100. Multidimensional stochastic processes as rough paths: theory and applications (2010)
  101. A course on rough paths (2020)
  102. A feature set for streams and an application to high-frequency financial tick data (2014)
  103. Derivatives pricing using signature payoffs (2018) — link
  104. Numerical method for model-free pricing of exotic derivatives in discrete time using rough path signatures (2019)
  105. Optimal execution with rough path signatures (2020)
  106. Chinese character-level writer identification using path signature feature, DropStroke and deep CNN (2015)
  107. Learning spatial-semantic context with fully convolutional recurrent network for online handwritten chinese text recognition (2017)
  108. LPSNet: a novel log path signature feature based hand gesture recognition framework (2017)
  109. Deep signature transforms (2019)
  110. Signatory: differentiable computations of the signature and logsignature transforms, on both CPU and GPU (2020)
  111. A generalised signature method for multivariate time series feature extraction (2020) — link

0 of 119 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.