Paper: arXiv 2407.13751

Abstract

In the era of rapid globalization and digitalization, accurate identification of similar stocks has become increasingly challenging due to the non-stationary nature of financial markets and the ambiguity in conventional regional and sector classifications. To address these challenges, we examine SimStock, a novel temporal self-supervised learning framework that combines techniques from self-supervised learning (SSL) and temporal domain generalization to learn robust and informative representations of financial time series data. The primary focus of our study is to understand the similarities between stocks from a broader perspective, considering the complex dynamics of the global financial landscape. We conduct extensive experiments on four real-world datasets with thousands of stocks and demonstrate the effectiveness of SimStock in finding similar stocks, outperforming existing methods. The practical utility of SimStock is showcased through its application to various investment strategies, such as pairs trading, index tracking, and portfolio optimization, where it leads to superior performance compared to conventional methods. Our findings empirically examine the potential of data-driven approach to enhance investment decision-making and risk management practices by leveraging the power of temporal self-supervised learning in the face of the ever-changing global financial landscape.

Complexity vs Empirical Score

  • Math Complexity: 7.5/10
  • Empirical Rigor: 8.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper uses advanced mathematical concepts including self-supervised learning and temporal domain generalization, requiring dense methodology, but also demonstrates high empirical rigor through extensive experiments on four real-world datasets with thousands of stocks and practical applications to investment strategies like pairs trading and portfolio optimization.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Identify similar stocks<br>under market non-stationarity"] --> B["Methodology: SimStock<br>Self-Supervised Learning +<br>Temporal Domain Generalization"]
  B --> C["Input Data<br>4 Real-world datasets<br>Thousands of stocks"]
  C --> D["Computational Process<br>Learn robust time-series<br>representations"]
  D --> E["Findings: Applications & Outcomes<br>Superior performance in<br>Pairs Trading, Index Tracking,<br>& Portfolio Optimization"]

Cited References (189)

  1. Portfolio Selection (1952) — link
  2. Xlnet: Generalized autoregressive pretraining for language understanding (2019)
  3. Transformer-based stagewise decomposition for large-scale multistage stochastic optimization (2023)
  4. Pairs trading: Performance of a relative-value arbitrage rule (2006)
  5. A methodology for index tracking based on time-series clustering (2004)
  6. Deep learning for portfolio optimization (2020) — link
  7. A survey on transfer learning (2009)
  8. Learning under concept drift: A review (2018)
and 181 more
  1. A comprehensive look at the empirical performance of equity premium prediction (2008)
  2. International asset allocation with regime shifts (2002)
  3. Asset allocation under multivariate regime switching (2007)
  4. Out-of-sample equity premium prediction: Combination forecasts and links to the real economy (2010)
  5. A fast algorithm for the minimum covariance determinant estimator (1999)
  6. Investing for the long run when returns are predictable (2000)
  7. Minimum volume ellipsoid (2009)
  8. A well-conditioned estimator for large-dimensional covariance matrices (2004)
  9. Improved estimation of the covariance matrix of stock returns with an application to portfolio selection (2003)
  10. Optimal estimation for economic gains: Portfolio choice with parameter uncertainty (2004)
  11. Optimal portfolio choice with parameter uncertainty (2007)
  12. Adaptive markets: Financial evolution at the speed of thought (2017)
  13. Empirical properties of asset returns: stylized facts and statistical issues (2001)
  14. A five-factor asset pricing model (2015)
  15. The valuation of risk assets and the selection of risky investments in stock portfolios and capital budgets (1975)
  16. Common risk factors in the returns on stocks and bonds (1993)
  17. Capital asset prices: A theory of market equilibrium under conditions of risk (1964)
  18. GRASP with path-relinking for the three-index assignment problem (2000)
  19. Mixtral of experts (2024) — link
  20. Self-supervised contrastive pre-training for time series via time-frequency consistency (2022)
  21. A simple framework for contrastive learning of visual representations (2020)
  22. Autoformer: Decomposition transformers with auto-correlation for long-term series forecasting (2021)
  23. Multi-task self-supervised time-series representation learning (2023) — link
  24. Fast RobustSTL: Efficient and robust seasonal-trend decomposition for time series with complex patterns (2020)
  25. Corrupted image modeling for self-supervised visual pre-training (2022) — link
  26. Patch-level representation learning for self-supervised vision transformers (2022)
  27. Emerging properties in self-supervised vision transformers (2021)
  28. Region similarity representation learning (2021)
  29. Time is matter: Temporal self-supervision for video transformers (2022) — link
  30. Masked feature prediction for self-supervised visual pre-training (2022)
  31. Momentum contrast for unsupervised visual representation learning (2020)
  32. Llama: Open and efficient foundation language models (2023) — link
  33. Language models are few-shot learners (2020) — link
  34. Bert: Pre-training of deep bidirectional transformers for language understanding (2018) — link
  35. Language models are unsupervised multitask learners (2019)
  36. Probability distribution of solution time in GRASP: an experimental investigation (2003)
  37. Parallel GRASP with path-relinking for job shop scheduling (1984)
  38. Title of a Book in Another Language (Quoted in that Language) (1988)
  39. Multi-start and strategic oscillation methods – principles to exploit adaptive memory (2000)
  40. Bubbles in Japan’s stock markets: a macroeconomic analysis (1995)
  41. The resurgent Japanese economy and a Japan–United States free trade agreement (1997)
  42. Efficient algorithms for layer assignment problems (1996)
  43. Gravitation (1973)
  44. Parallel GRASP with path-relinking for job shop scheduling (1983)
  45. Title of Book (2002)
  46. Does asymmetric information affect firm’s financing decisions? (2021)
  47. The role of asymmetric information on investments in emerging markets (2004)
  48. Lessons unlearned? Corporate debt in emerging markets (2017)
  49. Asymmetry of information and the finance-growth nexus in emerging markets: empirical evidence using panel VECM analysis (2012)
  50. Corporate debt, firm size and financial fragility in emerging markets (2019)
  51. Capital flows in risky times: Risk-on/risk-off and emerging market tail risk (2020)
  52. The dynamics of emerging market equity flows (2002)
  53. Cost of Asset Allocation in Equity Market: How Much Do Investors Lose Due to Bad Asset Class Design? (2014)
  54. An overview of machine learning for asset management (2023)
  55. Video representation learning by recognizing temporal transformations (2020)
  56. Spatiotemporal contrastive video representation learning (2021)
  57. Ts2vec: Towards universal representation of time series (2022)
  58. Domain randomization for transferring deep neural networks from simulation to the real world (2017)
  59. Training deep networks with synthetic data: Bridging the reality gap by domain randomization (2018)
  60. Domain-adversarial training of neural networks (2016)
  61. Dlow: Domain flow for adaptation and generalization (2019)
  62. Domain generalization and adaptation using low rank exemplar SVMs (2017)
  63. Domain generalization via model-agnostic learning of semantic features (2019)
  64. Temporal Domain Generalization with Drift-Aware Dynamic Neural Networks (2022) — link
  65. Best sources forward: domain generalization through source-specific nets (2018)
  66. Robust Detection of Lead-Lag Relationships in Lagged Multi-Factor Models — → 2305.06704
  67. Equity2vec: End-to-end deep learning framework for cross-sectional asset pricing (2021)
  68. Correlation Matrix Clustering for Statistical Arbitrage Portfolios (2023)
  69. Long short-term memory (1997)
  70. Exact indexing of dynamic time warping (2005)
  71. Bootstrap your own latent-a new approach to self-supervised learning (2020)
  72. Revisiting deep learning models for tabular data (2021)
  73. Attention is all you need (2017)
  74. Using dynamic time warping to find patterns in time series. (1994)
  75. Etsformer: Exponential smoothing transformers for time-series forecasting (2022) — link
  76. Fedformer: Frequency enhanced decomposed transformer for long-term series forecasting (2022)
  77. Exploring simple siamese representation learning (2021)
  78. Text and code embeddings by contrastive pre-training (2022) — link
  79. Autoformer: Decomposition transformers with auto-correlation for long-term series forecasting (2021)
  80. Learning local feature descriptors with triplets and shallow convolutional neural networks. (2016)
  81. Neural oblivious decision ensembles for deep learning on tabular data (2020)
  82. Scaling deep contrastive learning batch size under memory limited setup (2021) — link
  83. Barlow Twins: Self-Supervised Learning via Redundancy Reduction (2021) — link
  84. Understanding Contrastive Representation Learning through Alignment and Uniformity on the Hypersphere (2020) — link
  85. Multi-disease classification model using Strassen’s half of threshold (SHoT) training algorithm in healthcare sector (2021)
  86. Tabpfn: A transformer that solves small tabular classification problems in a second (2022) — link
  87. Balancing energy, latency and accuracy for mobile sensor data classification (2011)
  88. Neural Oblivious Decision Ensembles for Deep Learning on Tabular Data (2019) — link
  89. INVASE: Instance-wise Variable Selection using Neural Networks (2018)
  90. Self-Training With Noisy Student Improves ImageNet Classification (2020)
  91. Thematic Indexes: Expanding Dimensions of Indexing (2022)
  92. On the Impact of Thematic ETFs on Index Investing (2022)
  93. CutMix: Regularization Strategy to Train Strong Classifiers With Localizable Features (2019)
  94. AMLB: an AutoML Benchmark (2022) — link
  95. OpenML Benchmarking Suites (2021) — link
  96. Asymmetric Tri-training for Unsupervised Domain Adaptation (2017)
  97. Unsupervised Data Augmentation for Consistency Training (2020) — link
  98. Temporal Ensembling for Semi-Supervised Learning (2016)
  99. An Overview of Deep Semi-Supervised Learning (2020) — link
  100. AutoInt (2019) — link
  101. Adam: A Method for Stochastic Optimization (2015) — link
  102. Deep interest network for click-through rate prediction (2018)
  103. Adaptive energy management for self-sustainable wearables in mobile health (2022)
  104. Representation learning with contrastive predictive coding (2018) — link
  105. Feature selection model for healthcare analysis and classification using classifier ensemble technique (2021)
  106. Context-aware health event prediction via transition functions on dynamic disease graphs (2022)
  107. Statistical fraud detection: A review (2002)
  108. Building credit scoring models using genetic programming (2005)
  109. AutoGluon-Tabular: Robust and Accurate AutoML for Structured Data (2020) — link
  110. Identifying household finance heterogeneity via deep clustering (2022)
  111. Attention is All you Need (2017) — link
  112. The effect of errors in means, variances, and covariances on optimal portfolio choice (2013)
  113. Stop-loss adjusted labels for machine learning-based trading of risky assets (2023)
  114. The effects of errors in means, variances, and correlations on the mean-variance framework (2022)
  115. Similarity of neural network representations revisited (2019)
  116. Few-Shot Learning Patterns in Financial Time-Series for Trend-Following Strategies — → 2310.10500
  117. Language Guided Skill Discovery (2024) — link
  118. TopP&R: Robust Support Estimation Approach for Evaluating Fidelity and Diversity in Generative Models (2024)
  119. End-to-end risk budgeting portfolio optimization with neural networks (2023)
  120. A Survey of Large Language Models for Financial Applications: Progress, Prospects and Challenges — → 2406.11903
  121. The Markowitz optimization enigma: Is ‘optimized’optimal? (1989)
  122. Distributionally robust end-to-end portfolio construction (2023)
  123. Integrating prediction in mean-variance portfolio optimization (2023)
  124. A universal end-to-end approach to portfolio optimization via deep learning (2021) — link
  125. Enhancing time series momentum strategies using deep neural networks (2019) — link
  126. SimStock: Representation Model for Stock Similarities (2023)
  127. Sentiment correlation in financial news networks and associated market movements (2021)
  128. Honey, I Shrunk the Sample Covariance Matrix (2004)
  129. Portfolio selection (1952)
  130. Cointegration and error-correction models (2004)
  131. Trends and random walks in macroeconmic time series: some evidence and implications (1982)
  132. Deep Neural Networks and Tabular Data: A Survey (2022) — link
  133. Why do tree-based models still outperform deep learning on tabular data? (2022) — link
  134. Scarf: Self-supervised contrastive learning using random feature corruption (2021) — link
  135. On embeddings for numerical features in tabular deep learning (2022) — link
  136. Saint: Improved neural networks for tabular data via row attention and contrastive pre-training (2021) — link
  137. Tabnet: Attentive interpretable tabular learning (2021)
  138. Vime: Extending the success of self-and semi-supervised learning to tabular domain (2020)
  139. Tabtransformer: Tabular data modeling using contextual embeddings (2020) — link
  140. TaBERT: Pretraining for joint understanding of textual and tabular data (2020) — link
  141. Tabbie: Pretrained representations of tabular data (2021) — link
  142. Subtab: Subsetting features of tabular data for self-supervised representation learning (2021)
  143. Hopular: Modern Hopfield Networks for Tabular Data (2022) — link
  144. Self-attention between datapoints: Going beyond individual input-output pairs in deep learning (2021)
  145. TabGNN: Multiplex graph neural network for tabular data prediction (2021) — link
  146. TaPas: Weakly supervised table parsing via pre-training (2020) — link
  147. S4l: Self-supervised semi-supervised learning (2019)
  148. Self-supervised visual feature learning with deep neural networks: A survey (2020)
  149. Latent retrieval for weakly supervised open domain question answering (2019) — link
  150. Mpnet: Masked and permuted pre-training for language understanding (2020)
  151. Strong baselines for neural semi-supervised learning under domain shift (2018) — link
  152. Pre-trained models for natural language processing: A survey (2020)
  153. Imagenet: A large-scale hierarchical image database (2009)
  154. GLUE: A multi-task benchmark and analysis platform for natural language understanding (2018) — link
  155. Bagging predictors (1996)
  156. Evaluation of tree-based ensemble machine learning models in predicting stock price direction of movement (2020)
  157. Xgboost: A scalable tree boosting system (2016)
  158. Lightgbm: A highly efficient gradient boosting decision tree (2017)
  159. CatBoost: unbiased boosting with categorical features (2018)
  160. OpenML: networked science in machine learning (2014)
  161. Noise-contrastive estimation: A new estimation principle for unnormalized statistical models (2010)
  162. Understanding contrastive representation learning through alignment and uniformity on the hypersphere (2020)
  163. An open source AutoML benchmark (2019) — link
  164. Supervised and unsupervised discretization of continuous features (1995)
  165. Error-based and entropy-based discretization of continuous features. (1996)
  166. Cutmix: Regularization strategy to train strong classifiers with localizable features (2019)
  167. mixup: Beyond empirical risk minimization (2017) — link
  168. Temporal fusion transformers for interpretable multi-horizon time series forecasting (2021)
  169. INVASE: Instance-wise variable selection using neural networks (2019)
  170. Adam: A method for stochastic optimization (2014) — link
  171. Autogluon-tabular: Robust and accurate automl for structured data (2020) — link
  172. International Diversification—Still Not Crazy after All These Years (2023)
  173. Size, value, and momentum in international stock returns (2012)
  174. Value and momentum everywhere (2013)
  175. Size, value, and momentum in emerging market stock returns (2013)
  176. An artificial intelligence-enabled industry classification and its interpretation (2021)
  177. Estimation for Markowitz efficient portfolios (1980)
  178. Honey, I shrunk the sample covariance matrix (2003)
  179. The Gerber statistic: A robust co-movement measure for portfolio optimization (2021)
  180. Foundations of portfolio theory (1991)
  181. Mean-variance optimization for asset allocation (2021)

3 of 189 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.