Paper: arXiv 2406.08742

Abstract

This paper introduces DeepUnifiedMom, a deep learning framework that enhances portfolio management through a multi-task learning approach and a multi-gate mixture of experts. The essence of DeepUnifiedMom lies in its ability to create unified momentum portfolios that incorporate the dynamics of time series momentum across a spectrum of time frames, a feature often missing in traditional momentum strategies. Our comprehensive backtesting, encompassing diverse asset classes such as equity indexes, fixed income, foreign exchange, and commodities, demonstrates that DeepUnifiedMom consistently outperforms benchmark models, even after factoring in transaction costs. This superior performance underscores DeepUnifiedMom’s capability to capture the full spectrum of momentum opportunities within financial markets. The findings highlight DeepUnifiedMom as an effective tool for practitioners looking to exploit the entire range of momentum opportunities. It offers a compelling solution for improving risk-adjusted returns and is a valuable strategy for navigating the complexities of portfolio management.

Complexity vs Empirical Score

  • Math Complexity: 8.0/10
  • Empirical Rigor: 8.5/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper employs advanced deep learning architectures (multi-task learning with multi-gate mixture of experts) involving complex formulations, while demonstrating rigorous empirical validation with comprehensive backtests across multiple asset classes including transaction costs.

Research Flowchart

  flowchart TD
  A["Research Goal:<br>Create unified momentum portfolio<br>via Multi-task Learning & MoE"] --> B["Data & Inputs<br>Asset Classes: Equity, FX, Bonds, Commodities"]
  B --> C["Core Methodology<br>DeepUnifiedMom Framework"]
  C --> D["Multi-Task Learning<br>Long/Short signals across time horizons"]
  C --> E["Multi-Gate Mixture of Experts<br>Specialized expert networks"]
  D & E --> F["Computational Process<br>Unified Portfolio Construction"]
  F --> G["Backtesting & Evaluation<br>Transaction costs applied"]
  G --> H["Key Outcomes<br>Consistent outperformance<br>Superior risk-adjusted returns"]

Cited References (101)

  1. A Framework for Learning Predictive Structures from Multiple Tasks and Unlabeled Data (2005)
  2. Tied Multitask Learning for Neural Speech Translation (2018)
  3. A multimodal model with Twitter FinBERT embeddings for extreme price movement prediction of Bitcoin (2022) — link
  4. Forecasting Bitcoin volatility spikes from whale transactions and CryptoQuant data using Synthesizer Transformer models (2022) — link
  5. Fact, Fiction, and Momentum Investing (2014)
  6. Auxiliary task reweighting for minimum-data learning (2020)
  7. Task Clustering and Gating for Bayesian Multitask Learning (2003) — link
  8. Improving Time-Series Momentum Strategies: The Role of Trading Signals and Volatility Estimators (2012)
and 93 more
  1. Demystifying Time-Series Momentum Strategies: Volatility Estimators, Trading Rules and Pairwise Correlations (2021)
  2. Momentum has its moments (2015)
  3. A Model of Inductive Bias Learning (2000)
  4. Dissecting Investment Strategies in the Cross Section and Time Series (2015) — link
  5. Identifying beneficial task relations for multi-task learning in deep neural networks (2017)
  6. Global Portfolio Optimization (1992) — link
  7. Multitask Learning (1997)
  8. Drawdown Measure in Portfolio Optimization (2005) — link
  9. GradNorm: Gradient Normalization for Adaptive Loss Balancing in Deep Multitask Networks (2018)
  10. Portfolio Constraints and the Fundamental Law of Active Management (2002) — link
  11. Multi-task Learning Using Uncertainty to Weigh Losses for Scene Geometry and Semantics (2008)
  12. A unified architecture for natural language processing: deep neural networks with multitask learning (2008)
  13. Momentum crashes (2016)
  14. Optimal Versus Naive Diversification: How Inefficient is the 1/N Portfolio Strategy? (2007) — link
  15. Adapting Auxiliary Losses Using Gradient Similarity (2018)
  16. Low Resource Dependency Parsing: Cross-lingual Parameter Sharing in a Neural Network Parser (2015) — link
  17. Mean-Semivariance Optimization: A Heuristic Approach (2007) — link
  18. Efficient Capital Markets: A Review of Theory and Empirical Work (1970)
  19. Two Centuries of Price-Return Momentum (2016)
  20. The Trend Is Your Friend: Time-Series Momentum Strategies Across Equity and Commodity Markets (2016)
  21. Portfolio Selection (1952)
  22. A Century of Evidence on Trend-Following Investing (2017)
  23. Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency (1993)
  24. Profitability of Momentum Strategies: An Evaluation of Alternative Explanations (2001)
  25. Earnings Announcements are Full of Surprises (2008)
  26. Which Trend Is Your Friend? (2016)
  27. Auxiliary Tasks in Multi-task Learning (2018) — link
  28. Time series momentum (2012)
  29. A Survey on Transfer Learning (2010) — link
  30. The efficient market hypothesis and calendar anomalies: a literature review (2015)
  31. Enhancing the momentum strategy through deep regression (2019) — link
  32. An Overview of Multi-Task Learning in Deep Neural Networks (2017) — link
  33. The Sharpe Ratio (1994)
  34. Learning to Learn: Introduction and Overview (1998)
  35. Discovering Structure in Multiple Learning Tasks: The TC Algorithm (1996)
  36. Multi-Task Learning via Generalized Tensor Trace Norm (2021) — link
  37. TREND: Stevens Analytics. Continuous Futures: Future Daily Open Price | Symbol: – | Exchange: All Exchanges | Symbol: ED, 09/28/1983 - 10/23/2020. Data Planet™ Statistical Datasets: A SAGE Publishing Resource Dataset-ID: 095-001-006 (2020) — link
  38. Time Series Prediction Using Support Vector Machines: A Survey (2009)
  39. Practical Bayesian support vector regression for financial time series prediction and market condition change detection (2017) — link
  40. Application of support vector machines in financial time series forecasting (2001) — link
  41. Financial time series forecasting using support vector machines (2003) — link
  42. Support vector machine with adaptive parameters in financial time series forecasting (2003) — link
  43. Long Short-Term Memory (1997) — link
  44. The Extreme Value Method for Estimating the Variance of the Rate of Return (1980)
  45. On the Estimation of Security Price Volatilities from Historical Data (1980)
  46. Estimating Variance From High, Low and Closing Prices (1991)
  47. Drift‐Independent Volatility Estimation Based on High, Low, Open, and Close Prices (2000)
  48. Enhancing Time Series Momentum Strategies Using Deep Neural Networks (2019) — link
  49. Trading with the Momentum Transformer: An Intelligent and Interpretable Architecture (2021) — link
  50. Attention is All you Need (2017)
  51. Deep learning for portfolio optimisation (2020) — link
  52. Trading with the Momentum Transformer: An intelligent and interpretable architecture (2022) — link
  53. COVID-19 and the march 2020 stock market crash. Evidence from S&P1500 (2021) — link
  54. The 60/40 problem: Examining the traditional 60/40 portfolio in an uncertain rate environment (2017) — link
  55. We see risk where others may not: 60/40 in 2020 vision (2020) — link
  56. Multi-Task Learning for Stock Selection (1996) — link
  57. A comparison of multitask and single task learning with artificial neural networks for yield curve forecasting (2019) — link
  58. Federated Reinforcement Learning for Portfolio Management (2022) — link
  59. Strategic risk management: Designing portfolios and managing risk (2021)
  60. Continuous control with Stacked Deep Dynamic Recurrent Reinforcement Learning for portfolio optimization (2020) — link
  61. A CWGAN-GP-based multi-task learning model for consumer credit scoring (2022) — link
  62. COVID19-MLSF: A multi-task learning-based stock market forecasting framework during the COVID-19 pandemic (2023) — link
  63. Stock Ranking with Multi-Task Learning (2022) — link
  64. Constructing time-series momentum portfolios with deep multi-task learning — → 2306.13661
  65. Strategic Risk Management: Out-of-Sample Evidence from the COVID-19 Equity Selloff (2020)
  66. Network Momentum across Asset Classes (2023) — link
  67. Performance of Time-series Momentum Strategy: US Evidence (2023) — link
  68. Designing Robust Trend-following System: Behind the Scenes of Trend-following (2018)
  69. Long-Only Multi-Asset Momentum: Searching for Absolute Returns (2022)
  70. Uncovering Momentum (2020)
  71. Momentum Turning Points (2023)
  72. Trend Following - Expected Returns (2015)
  73. Adaptive Mixtures of Local Experts (1991) — link
  74. Modeling Task Relationships in Multi-task Learning with Multi-gate Mixture-of-Experts (2018) — link
  75. Switch transformers: scaling to trillion parameter models with simple and efficient sparsity (2022)
  76. Mixture-of-Experts Meets Instruction Tuning:A Winning Combination for Large Language Models (2023) — link
  77. ST-MoE: Designing Stable and Transferable Sparse Expert Models (2022) — link
  78. FasterMoE: modeling and optimizing training of large-scale dynamic pre-trained models (2022) — link
  79. MegaBlocks: Efficient Sparse Training with Mixture-of-Experts (2022) — link
  80. Estimation Error and Portfolio Optimization: A Resampling Solution (2007)
  81. Active Risk-Based Investing (2015) — link
  82. The Sharpe ratio efficient frontier (2012) — link
  83. Building Diversified Portfolios that Outperform Out of Sample: (2016) — link
  84. Advances in Financial Machine Learning (2018)
  85. Can Machines “Learn” Finance? (2020) — link
  86. The Probability of Backtest Overfitting (2015)
  87. Evaluating Trading Strategies (2014)
  88. Estimating the significance of a signal in a multi-dimensional search (2011) — link
  89. Deep Learning-based Integrated Framework for stock price movement prediction (2023) — link
  90. Deep learning for financial applications : A survey (2020) — link
  91. Research on the construction of stock portfolios based on multiobjective water cycle algorithm and KMV algorithm (2022) — link
  92. Robust mean-risk portfolio optimization using machine learning-based trade-off parameter (2021) — link
  93. Multiagent-based deep reinforcement learning for risk-shifting portfolio management (2022) — link

1 of 101 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.