---
title: A Variational Autoencoder Approach to Conditional Generation of Possible Future Volatility Surfaces
url: https://www.ml-quant.com/papers/ssrn/4628457/
site: ML-Quant (https://www.ml-quant.com)
updated: 2026-09-26
license: Summaries CC BY 4.0; links go to the original sources
index: https://www.ml-quant.com/llms.txt
identifier: SSRN 4628457
source_url: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4628457
featured: 2023-11-15
citations: 2
topic: Derivatives & Volatility
---


# A Variational Autoencoder Approach to Conditional Generation of Possible Future Volatility Surfaces

The paper presents a new method for predicting future implied volatility surfaces using historical data, employing a conditional variational autoencoder and a long short-term memory network.

- Source: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4628457
- Identifier: SSRN 4628457
- Released: 2023-11-09
- First featured: Quant Letter No. 26 (2023-11-15): https://www.ml-quant.com/issues/2023-11-15/
- Citations (Semantic Scholar): 2
- Published in: not yet
- Topic: Derivatives & Volatility

## Related

- [Computing Volatility Surfaces using Generative Adversarial Networks with Minimal Arbitrage Violations](https://www.ml-quant.com/papers/arxiv/2304.13128/): The article introduces a generative adversarial network (GAN) method for calculating volatility surfaces, showing that the GAN model is more accurate and faster than artificial neural network (ANN) methods.
- [Filling in Missing FX Implied Volatilities with Uncertainties: Improving VAE-Based Volatility Imputation](https://www.ml-quant.com/papers/arxiv/2411.05998/): The study explores enhancing the prediction of missing implied volatilities in FX options using modified variational autoencoders (VAEs), which better manage data uncertainty.
- [Universal Diffusion Models for Implied Volatility Surfaces: Learning Shared Dynamics Across Stocks](https://www.ml-quant.com/papers/arxiv/2609.22893/): A universal diffusion model trained on pooled data from 50 stocks learns to jointly generate implied volatility surface changes and stock returns, extrapolating well to unseen stocks.
- [A Gaussian Process Based Method with Deep Kernel Learning for Pricing High-Dimensional American Options](https://www.ml-quant.com/papers/arxiv/2311.07211/): Deep Kernel Learning and variational inference are used to improve high-dimensional American option pricing in the regression-based Monte Carlo method, with successful performance under geometric Brownian motion and Merton's jump diffusion models.
- [Volatility Transformers: an optimal transport-inspired approach to arbitrage-free shaping of implied volatility surfaces](https://www.ml-quant.com/papers/ssrn/4623940/): Arbitrage-Free Volatility Surfaces: The paper presents a framework for creating arbitrage-free transformations of an implied volatility surface using optimal transport maps, which can be applied to a broader range of synthetic market data generation applications.
- [Power law in Sandwiched Volterra Volatility model](https://www.ml-quant.com/papers/arxiv/2311.01228/): Power Law in Volterra Volatility Model: The Sandwiched Volterra Volatility (SVV) model accurately reproduces the power-law behavior of the at-the-money implied volatility skew, provided the correct Volterra kernel is chosen.