Skip to content
DraconiticResearch
Search Sources API
181,716 studies31 sourcesUpdated 40 min ago
PreprintarXiv licencearXiv

Estimating "Realized" Skewness using Convolutional Neural Network

  • Haozhe Jiang
  • Ostap Okhrin
  • Michael Rockinger
Open at sourcePDFDOICite

Abstract

We propose a new estimator of low-frequency skewness (monthly) that exploits high-frequency data (5 min) through a direct functional mapping consisting of layers of Convolutional Neural Networks followed by Multi Layer Perceptrons. Monte Carlo experiments involving stochastic volatility models, with and without jumps, show that the estimator reduces finite-sample bias relative to existing realized-skewness estimators and remains accurate under model misspecification. Empirically, our estimator exhibits temporal stability and delivers cross-sectional performance in actual skewness-sorted portfolios. We characterize the finite-sample bias of conventional plug-in skewness estimators and establish risk consistency properties of the proposed NN estimator under standard regularity conditions.

Related

  1. Stochastic dynamics generation of directed percolation using a two-channel convolutional neural network
    Feng Gao, Jianmin Shen, Wei Li
    arXivPreprintarXiv licenceFull textMathematical physicsarXiv
  2. Initial condition recovery in nonlinear damped viscous photoacoustic tomography using a convolutional neural network-guided gradient-free optimization framework
    Madhu Gupta, Anwesa Dey, Prapti Tala +1
    arXivPreprintCC BY 4.0Full textOptimisation & controlarXiv
  3. Memorization Capacity of Invariant Convolutional Neural Nets
    Sjoerd Dirksen, Patrick Finke, Paul Geuchen
    arXivPreprintCC BY 4.0Full textStatisticsarXiv
  4. The role of locality and weight sharing in convolutional neural networks
    Francesco Mori, Melanie Weber
    arXivPreprintCC BY 4.0Full textStatisticsCondensed matterarXiv
  5. Estimating Causal Bounds from Observational Data With Neural Generative Models
    Max Zhu, Srivathsan Amruth, Sina Akbari +1
    arXivPreprintCC BY 4.0Full textStatisticsarXiv

Cite

BibTeXRIS
BibTeX
@misc{jiang2026estimating,
  title = {{Estimating "Realized" Skewness using Convolutional Neural Network}},
  author = {Haozhe Jiang and Ostap Okhrin and Michael Rockinger},
  howpublished = {arXiv},
  year = {2026},
  month = nov,
  doi = {10.48550/arxiv.2611.04598},
  eprint = {2611.04598},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2611.04598},
}
RIS
TY  - UNPB
TI  - Estimating "Realized" Skewness using Convolutional Neural Network
AU  - Haozhe Jiang
AU  - Ostap Okhrin
AU  - Michael Rockinger
PY  - 2026
DA  - 2026/11/03
PB  - arXiv
DO  - 10.48550/arxiv.2611.04598
UR  - https://arxiv.org/abs/2611.04598
AB  - We propose a new estimator of low-frequency skewness (monthly) that exploits high-frequency data (5 min) through a direct functional mapping consisting of layers of Convolutional Neural Networks followed by Multi Layer Perceptrons. Monte Carlo experiments involving stochastic volatility models, with and without jumps, show that the estimator reduces finite-sample bias relative to existing realized-skewness estimators and remains accurate under model misspecification. Empirically, our estimator exhibits temporal stability and delivers cross-sectional performance in actual skewness-sorted portfolios. We characterize the finite-sample bias of conventional plug-in skewness estimators and establish risk consistency properties of the proposed NN estimator under standard regularity conditions.
ER  -