Skip to content

Volatility Surface Reconstruction using Deep Learning under No-Arbitrage Constraints

Bibliographic record. Follow the original-source link for the publication.

Field Value
Primary domain Volatility
Other domains
Methods Financial Ml
Facets
Authors Pablo Rodriguez Manzi
Published 2026-05-20
Source arXiv Quantitative Finance History
Identifiers arxiv:2605.24031
URL Open original source

Editorial synthesis

Why it matters

This study reconstructs implied-volatility surfaces from sparse noisy quotes with no-arbitrage constraints, which is important for quote-clean environments and model selection under incomplete markets. (abstract:S1, abstract:S2, abstract:S3, abstract:S4)

Main author claims

  • The authors claim to reconstruct implied-volatility surfaces under no-arbitrage constraints from sparse and noisy quotes and benchmark multiple network architectures against classical SVI parameterizations. (abstract:S1, abstract:S2)
  • They further claim Transformer and U-Net perform strongly, especially in sparse regimes, and that soft arbitrage penalties reduce arbitrage violations with moderate effect on reconstruction error. (abstract:S3, abstract:S4)

Data, method, or discussion scope

The scope includes sparse-noisy quote reconstruction, architecture set, sparse-regime performance claims, and explicit accuracy versus arbitrage-consistency trade-off analysis. (abstract:S1, abstract:S2, abstract:S3, abstract:S4)

Main limitations

Exact regularization grids, compute costs, and out-of-domain error bounds are absent, as are operational metrics for risk and latency. (abstract:S3, abstract:S4)

Relationships

  • None recorded.