Paper: arXiv 2402.06642

Abstract

Volatility, as a measure of uncertainty, plays a crucial role in numerous financial activities such as risk management. The Econometrics and Machine Learning communities have developed two distinct approaches for financial volatility forecasting: the stochastic approach and the neural network (NN) approach. Despite their individual strengths, these methodologies have conventionally evolved in separate research trajectories with little interaction between them. This study endeavors to bridge this gap by establishing an equivalence relationship between models of the GARCH family and their corresponding NN counterparts. With the equivalence relationship established, we introduce an innovative approach, named GARCH-NN, for constructing NN-based volatility models. It obtains the NN counterparts of GARCH models and integrates them as components into an established NN architecture, thereby seamlessly infusing volatility stylized facts (SFs) inherent in the GARCH models into the neural network. We develop the GARCH-LSTM model to showcase the power of the GARCH-NN approach. Experiment results validate that amalgamating the NN counterparts of the GARCH family models into established NN models leads to enhanced outcomes compared to employing the stochastic and NN models in isolation.

Complexity vs Empirical Score

  • Math Complexity: 6.5/10
  • Empirical Rigor: 7.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper introduces significant mathematical complexity by establishing equivalence relationships between GARCH models and neural network counterparts, requiring advanced econometric and machine learning theory. Empirical rigor is demonstrated through experiments on five globally traded equities datasets and Value at Risk (VaR) computation, showing backtest-ready implementation with specific performance comparisons.

Research Flowchart

  flowchart TD
  A["Research Goal: Bridge Gap<br>between GARCH and NN<br>for Volatility Forecasting"] --> B["Key Methodology: Establish<br>Equivalence Relationship<br>between GARCH & NN Models"]
  B --> C["Develop GARCH-NN Approach<br>Integrate GARCH NN-counterparts<br>into NN Architecture"]
  C --> D{"Computational Process"}
  D --> E["Data Input: Financial Time Series"]
  D --> F["Model 1: Stochastic Approach<br>e.g., Standard GARCH"]
  D --> G["Model 2: Pure NN Approach<br>e.g., LSTM"]
  D --> H["Model 3: Proposed Method<br>GARCH-LSTM"]
  E --> F & G & H
  F & G & H --> I["Key Outcomes: Enhanced<br>Forecasting Performance<br>via Hybrid Architecture"]

Cited References (124)

  1. bashtage/arch: Release 6.1.0 (2023) — link
  2. A hybrid method of exponential smoothing and recurrent neural networks for time series forecasting (2020)
  3. A Framework for Stochastic and Surrogate-Assisted Optimization using Sequential Modular Process Simulators (2020)
  4. Do We Really Need Deep Learning Models for Time Series Forecasting? (2021) — link
  5. Are Transformers Effective for Time Series Forecasting? (2022) — link
  6. Statistical and Machine Learning forecasting methods: Concerns and ways forward (2018) — link
  7. A hybrid deep learning approach by integrating LSTM-ANN networks with GARCH model for copper price volatility prediction (2020) — link
  8. Value-at-Risk forecasting: A hybrid ensemble learning GARCH-LSTM based approach (2022)
and 116 more
  1. On the relationship of implied, realized and historical volatility: evidence from NSE equity index options (2014)
  2. Value at Risk (VaR) Using Volatility Forecasting Models: EWMA, GARCH and Stochastic Volatility (2007)
  3. bashtage/arch: Release 5.3.1 (2022) — link
  4. Multivariate GARCH models: a survey (2006)
  5. Noisy Time Series Prediction using Recurrent Neural Networks and Grammatical Inference (2001)
  6. HTML: Hierarchical Transformer-based Multi-task Learning for Volatility Prediction (2020)
  7. A Survey on Explainable Artificial Intelligence ( XAI ): Toward Medical XAI (2021)
  8. A forecast comparison of volatility models: does anything beat a GARCH(1,1)? (2005)
  9. Forecasting volatility of oil price using an artificial neural network-GARCH model (2016)
  10. Exploiting the low-risk anomaly using machine learning to enhance the Black–Litterman framework: Evidence from South Korea (2018)
  11. Determining the relationship between speculative activity and crude oil price volatility, using artificial neural networks (2017)
  12. Neural Network–Based Financial Volatility Forecasting: A Systematic Review (2022)
  13. Statistical and Machine Learning forecasting methods: Concerns and ways forward (2018)
  14. Volatility Stylized Facts (2011) — link
  15. Long memory and nonlinearity in conditional variances: A smooth transition FIGARCH model (2011)
  16. N-hits: Neural hierarchical interpolation for time series forecasting (2022) — link
  17. Enhancing the locality and breaking the memory bottleneck of transformer on time series forecasting (2019)
  18. Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting (2021)
  19. Reformer: The efficient transformer (2020) — link
  20. DeepAR: Probabilistic forecasting with autoregressive recurrent networks (2020)
  21. N-BEATS: Neural basis expansion analysis for interpretable time series forecasting (2019) — link
  22. Think globally, act locally: A deep neural network approach to high-dimensional time series forecasting (2019)
  23. Modeling long-and short-term temporal patterns with deep neural networks (2018)
  24. Autoformer: Decomposition transformers with auto-correlation for long-term series forecasting (2021)
  25. Volatility Clustering in Financial Markets: Empirical Facts and Agent-Based Models (2007)
  26. Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom inflation (1982)
  27. Generalized autoregressive conditional heteroskedasticity (1986)
  28. A Conditionally Heteroskedastic Time Series Model for Speculative Prices and Rates of Return (1987) — link
  29. On the Relation between the Expected Value and the Volatility of the Nominal Excess Return on Stocks (1993)
  30. Fractionally integrated generalized autoregressive conditional heteroskedasticity (1996)
  31. Moment and Memory Properties of Linear Conditional Heteroscedasticity Models, and a New Model (2004)
  32. Tsmixer: An all-mlp architecture for time series forecasting (2023) — link
  33. Gold price volatility: A forecasting approach using the Artificial Neural Network-GARCH model (2015)
  34. Forecasting the volatility of stock price index: A hybrid model integrating LSTM with multiple GARCH-type models (2018)
  35. Conditional Heteroskedasticity in Asset Returns: A New Approach (1991)
  36. Approximation by superpositions of a sigmoidal function (1989)
  37. Volatility Forecast Using Hybrid Neural Network Models (2014)
  38. Machine learning for realised volatility forecasting (2020)
  39. Improving GARCH volatility forecasts with regime-switching GARCH (2002)
  40. Multi-transformer: A new neural network-based architecture for forecasting S&P volatility (2021)
  41. Attention is All you Need (2017)
  42. Stock Volatility Prediction Based on Self-attention Networks with Social Information (2019)
  43. A Simple Approximate Long-Memory Model of Realized Volatility (2009)
  44. GARCH Parameter Estimation by Machine Learning (2018)
  45. A Novel Connectionist System for Unconstrained Handwriting Recognition. (2009)
  46. On Mixture Memory Garch Models (2013)
  47. Longer-Term Time-Series Volatility Forecasts (2010)
  48. Moment and Memory Properties of Linear Conditional Heteroscedasticity Models, and a New Model (2004) — link
  49. Forecasting volatility (2005)
  50. Market Risk Prediction under Long Memory: When VaR is Higher than Expected (2012) — link
  51. On conditional moments of GARCH models, with applications to multiple period value at risk estimation (2003)
  52. Explainable Deep Neural Networks for Multivariate Time Series Predictions — link
  53. Explainable Artificial Intelligence ( XAI ) on TimeSeries Data: A Survey — link
  54. Evaluating Explainable AI : Which Algorithmic Explanations Help Users Predict Model Behavior? — link
  55. A deep learning framework for financial time series using stacked autoencoders and long-short term memory — link
  56. ST -Trader: A Spatial-Temporal Deep Neural Network for Modeling Stock Market Movement — link
  57. A graph-based convolutional neural network stock price prediction with leading indicators — link
  58. Stock index prediction and uncertainty analysis using multi-scale nonlinear ensemble paradigm of optimal feature extraction, two-stage deep learning and Gaussian process regression — link
  59. A hybrid approach of adaptive wavelet transform, long short-term memory and ARIMA - GARCH family models for the stock index prediction — link
  60. A Volume-Aware Positional Attention-Based Recurrent Neural Network for Stock Index Prediction — link
  61. Stock price forecasting model based on modified convolution neural network and financial time series analysis — link
  62. Exploration of stock index change prediction model based on the combination of principal component analysis and artificial neural network — link
  63. The Application of Stock Index Price Prediction with Neural Network — link
  64. An improved Stacking framework for stock index prediction by leveraging tree-based ensemble models and deep learning algorithms — link
  65. How is Machine Learning Useful for Macroeconomic Forecasting? — link
  66. Macroeconomic forecasting using factor models and machine learning: an application to Japan — link
  67. A Unified Approach to Interpreting Model Predictions — link
  68. Migration motif: a spatial - temporal pattern mining approach for financial markets — link
  69. Deep hedging — link
  70. Deep learning for financial applications : A survey — link
  71. A top-down approach to identifying bull and bear market states — link
  72. Structural Changes in the Duration of Bull Markets and Business Cycle Dynamics — link
  73. Deep Learning for Time-Series Analysis — link
  74. Improving GARCH volatility forecasts with regime-switching GARCH — link
  75. Volatility forecasts, proxies and loss functions — link
  76. Volatility forecast comparison using imperfect volatility proxies — link
  77. Volatility Forecasting Performance of GARCH Models:
  78. Predicting the Volatility of Cryptocurrency Time-Series — link
  79. Oil price volatility forecast with mixture memory GARCH — link
  80. Volatility Forecasting With Range-Based EGARCH Models — link
  81. Volatility forecasting and risk management for commodity markets in the presence of asymmetry and long memory — link
  82. Improving Pairs Trading Strategies via Reinforcement Learning — link
  83. Evaluation of the Transformer Architecture for Univariate Time Series Forecasting — link
  84. A Transformer Self-attention Model for Time Series Forecasting — link
  85. Volatility dynamics of crypto-currencies’ returns: Evidence from asymmetric and long memory GARCH models — link
  86. An Introduction to Univariate GARCH Models — link
  87. Financial time series forecasting with deep learning : A systematic literature review: 2005–2019 — link
  88. Bitcoin Network Mechanics: Forecasting the BTC Closing Price Using Vector Auto-Regression Models Based on Endogenous and Exogenous Feature Variables — link
  89. Temporal Attention-Augmented Bilinear Network for Financial Time-Series Data Analysis — link
  90. Temporal Fusion Transformers for interpretable multi-horizon time series forecasting — link
  91. Fuzzy time series forecasting based on optimal partitions of intervals and optimal weighting vectors — link
  92. Conditional Time Series Forecasting with Convolutional Neural Networks — link
  93. Multi-Horizon Time Series Forecasting with Temporal Attention Learning — link
  94. A deep asymmetric Laplace neural network for deterministic and probabilistic wind power forecasting — link
  95. Financial news-based stock movement prediction using causality analysis of influence in the Korean stock market — link
  96. Connecting the Dots: Multivariate Time Series Forecasting with Graph Neural Networks — link
  97. Transductive LSTM for time-series prediction: An application to weather forecasting — link
  98. Deep Learning with Long Short-Term Memory for Time Series Prediction — link
  99. Multiple-output modeling for multi-step-ahead time series forecasting — link
  100. Cryptocurrency volatility forecasting: A Markov regime-switching MIDAS approach — link
  101. Coherent Probabilistic Aggregate Queries on Long-horizon Forecasts — link
  102. Enhancing the Locality and Breaking the Memory Bottleneck of Transformer on Time Series Forecasting — link
  103. DeepAR : Probabilistic forecasting with autoregressive recurrent networks — link
  104. Forecasting at Scale — link
  105. N-BEATS: Neural basis expansion analysis for interpretable time series forecasting (2019) — link
  106. Attention is all you need (2017)
  107. Realized volatility forecasting with neural networks (2020)
  108. Forecasting realized volatility: a review (2017)
  109. Modeling and forecasting realized volatility (2003)
  110. Econometric analysis (2003)
  111. A software package for sequential quadratic programming (1988)
  112. A machine learning approach to volatility forecasting (2021)
  113. Using machine learning to predict realized variance (2019) — link
  114. The infinite Gaussian mixture model (1999)
  115. Ensemble deep learning: A review (2022)
  116. Asymmetric long memory GARCH in exchange return (2001)

0 of 124 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.