Paper: arXiv 2402.14476

Abstract

Time-series with volatility clustering pose a unique challenge to uncertainty quantification (UQ) for returns forecasts. Methods for UQ such as Deep Evidential regression offer a simple way of quantifying return forecast uncertainty without the costs of a full Bayesian treatment. However, the Normal-Inverse-Gamma (NIG) prior adopted by Deep Evidential regression is prone to miscalibration as the NIG prior is assigned to latent mean and variance parameters in a hierarchical structure. Moreover, it also overparameterizes the marginal data distribution. These limitations may affect the accurate delineation of epistemic (model) and aleatoric (data) uncertainties. We propose a Scale Mixture Distribution as a simpler alternative which can provide favorable complexity-accuracy trade-off and assign separate subnetworks to each model parameter. To illustrate the performance of our proposed method, we apply it to two sets of financial time-series exhibiting volatility clustering: cryptocurrencies and U.S. equities and test the performance in some ablation studies.

Complexity vs Empirical Score

  • Math Complexity: 8.5/10
  • Empirical Rigor: 7.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper introduces a novel Scale Mixture Distribution for uncertainty quantification, involving advanced hierarchical probability theory and the interplay between aleatoric and epistemic uncertainties (high math). It validates the method with concrete backtests on financial datasets (cryptocurrencies and U.S. equities) and performs ablation studies, demonstrating implementation-heavy empirical rigor.

Research Flowchart

  flowchart TD
  A["Research Goal:<br>Quantify Neural Network<br>Uncertainty under<br>Volatility Clustering"] --> B{"Identify Limitations<br>of Deep Evidential Regression"}
  B --> C["Proposed Method:<br>Scale Mixture Distribution"]
  C --> D["Data & Experiments:<br>Cryptos & US Equities<br>+ Ablation Studies"]
  D --> E["Computational Process:<br>Separate Subnetworks for<br>Model Parameters"]
  E --> F["Key Findings:<br>Favorable Complexity-Accuracy<br>Trade-off & Better UQ"]

Cited References (95)

  1. End to End Learning for Self-Driving Cars (2016) — link
  2. Deep Learning (2016)
  3. Deep Evidential Regression (2020)
  4. Language Models are Few-Shot Learners (2020)
  5. Advances in Neural Information Processing Systems 33 (2020)
  6. Supervised Temporal Autoencoder for Stock Return Time-series Forecasting (2021)
  7. Proceedings of the IEEE 45th Annual Computer Software and Applications Conference ( COMPSAC ‘21) (2021)
  8. Bayesian theory (2000)
and 87 more
  1. The Exponential Family: Conjugate Priors (2009)
  2. Convex optimization (2004)
  3. Simple and Scalable Predictive Uncertainty Estimation Using Deep Ensembles (2017)
  4. What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision? (2017)
  5. Advances in Neural Information Processing Systems 30 (2017)
  6. Bagging Predictors (1996)
  7. Online Prediction Under Model Uncertainty via Dynamic Model Averaging: Application to a Cold Rolling Mill (2010) — link
  8. Empirical properties of asset returns: stylized facts and statistical issues (2001)
  9. A practical guide to volatility forecasting through calm and storm (2011)
  10. Probabilistic Backpropagation for Scalable Learning of Bayesian Neural Networks (2015)
  11. Proceedings of the 32nd International Conference on Machine Learning (ICML) (2015)
  12. Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning (2016)
  13. Proceedings of the 33rd International Conference on Machine Learning (ICML) (2016)
  14. Cryptocurrencies change everything (2021) — link
  15. Investing with cryptocurrencies – evaluating their potential for portfolio allocation strategies (2021)
  16. Negative bubbles and shocks in cryptocurrency markets (2016)
  17. Testing for Bubbles in Cryptocurrencies with Time-Varying Volatility (2018) — link
  18. Sentiment-Induced Bubbles in the Cryptocurrency Market (2019)
  19. Statistical analysis of bitcoin during explosive behavior periods (2019) — link
  20. Profit Potential in an Almost Efficient Market (1974)
  21. Active Portfolio Management: A Quantitative Approach for Producing Superior Returns and Controlling Risk (1999)
  22. Pattern Recognition and Machine Learning (2006)
  23. A New Interpretation of Information Rate (1956)
  24. Portfolio Selection (1952)
  25. Asset Allocation: Combining Investor Views with Market Equilibrium (1991)
  26. A Survey of Uncertainty in Deep Neural Networks (2021)
  27. The Elements of Statistical Learning (2020)
  28. Neural Network–Based Financial Volatility Forecasting: A Systematic Review (2022)
  29. Bayesian Learning for Neural Networks (1996)
  30. Predictability of Stock Returns: Robustness and Economic Significance (1995)
  31. Long Short-Term Memory (1997)
  32. Autoregressive Conditional Heteroscedasticity with Estimates of the Variance of United Kingdom Inflation (1982)
  33. Generalized autoregressive conditional heteroskedasticity (1986) — link
  34. Time Series Analysis: Forecasting and Control (1994)
  35. A Comparison of ARIMA and LSTM in Forecasting Time Series (2018)
  36. Proceedings of the 17th IEEE International Conference on Machine Learning and Applications (2018)
  37. A Large-Scale Study of Probabilistic Calibration in Neural Network Regression (2023)
  38. Uncertainty Estimation by Fisher Information-based Evidential Deep Learning (2023)
  39. Proceedings of the 40th International Conference on Machine Learning (ICML) (2023)
  40. Novel volatility forecasting using deep learning–Long Short Term Memory Recurrent Neural Networks (2019)
  41. Generalized Framework For Applying The Kelly Criterion To Stock Markets (2018) — link
  42. Aleatoric and epistemic uncertainty in machine learning: an introduction to concepts and methods (2021)
  43. A Practical Bayesian Framework for Backpropagation Networks (1992) — link
  44. Uncertainty in Deep Learning (2016)
  45. Monte Carlo Sampling Methods Using Markov Chains and Their Applications (1970)
  46. On the Validity of Bayesian Neural Networks for Uncertainty Estimation (2019)
  47. Being Bayesian, Even Just a Bit, Fixes Overconfidence in ReLU Networks (2020)
  48. Can You Trust Your Model’s Uncertainty? Evaluating Predictive Uncertainty under Dataset Shift (2019)
  49. Advances in Neural Information Processing Systems 32 (2019)
  50. Speeding Up MCMC by Efficient Data Subsampling (2019) — link
  51. Scale mixtures of normal distributions (1974)
  52. Scale Mixtures Distributions in Statistical Modelling (2008)
  53. Empirical Asset Pricing via Machine Learning (2020) — link
  54. WaveNet: a deep convolutional neural network using directional wavelets for low-dose X-ray CT reconstruction (2017) — link
  55. Hands-On Bayesian Neural Networks — A Tutorial for Deep Learning Users (2022)
  56. The Unreasonable Effectiveness of Deep Evidential Regression (2022)
  57. Efficient and Robust LiDAR-Based End-to-End Navigation (2021)
  58. 2021 IEEE International Conference on Robotics and Automation (2021)
  59. Evidential Deep Learning for Guided Molecular Property Prediction and Discovery (2021)
  60. Vision-Based Autonomous Car Racing Using Deep Imitative Reinforcement Learning (2021) — link
  61. Leveraging Graph and Deep Learning Uncertainties to Detect Anomalous Maritime Trajectories (2022) — link
  62. 3D High-Quality Magnetic Resonance Image Restoration in Clinics Using Deep Learning (2022)
  63. On Second-Order Scoring Rules for Epistemic Uncertainty Quantification (2023)
  64. Sources of Uncertainty in Machine Learning – A Statisticians’ View (2023)
  65. Introduction to Uncertainty Quantification (2015)
  66. Aleatory and epistemic uncertainty in probability elicitation with an example from hazardous waste management: Treatment of aleatory and epistemic uncertainty (1996)
  67. The Well-Calibrated Bayesian (1982)
  68. The Comparison and Evaluation of Forecasters (1983)
  69. Lessons Learned in the Challenge: Making Predictions and Scoring Them (2006)
  70. Improving Evidential Deep Learning via Multi-Task Learning (2022)
  71. Natural Posterior Network: Deep Bayesian Predictive Uncertainty for Exponential Family Distributions (2021)
  72. 9th International Conference on Learning Representations (ICLR) (2021)
  73. Regression Prior Networks (2020)
  74. Strictly Proper Scoring Rules, Prediction, and Estimation (2007)
  75. GARCH Modeling of Stock Market Volatility
  76. Stock market volatility (2009)
  77. Adam: A Method for Stochastic Optimization (2015)
  78. 3rd International Conference on Learning Representations (2015)
  79. Generalized autoregressive conditional heteroskedasticity (1986)
  80. Mixture Density Networks (1994) — link
  81. Multivariate Deep Evidential Regression (2022) — link
  82. Uncertainty Regularized Evidential Regression (2024) — link
  83. Deep Evidential Learning for Bayesian Quantile Regression (2023) — link
  84. A Survey on Uncertainty Quantification Methods for Deep Learning (2024) — link
  85. A survey on uncertainty quantification in deep learning for financial time series prediction (2024) — link
  86. Analyzing return asymmetry and quantiles through stochastic volatility models using asymmetric Laplace error via uniform scale mixtures (2015)
  87. A new look at cryptocurrencies (2018)

0 of 95 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.