English

How to build a cross-impact model from first principles: Theoretical requirements and empirical results

Trading and Market Microstructure 2022-03-30 v3 Statistical Mechanics

Abstract

Trading a financial instrument pushes its price and those of other assets, a phenomenon known as cross-impact. To be of use, cross-impact models must fit data and be well-behaved so they can be applied in applications such as optimal trading. To address these issues, we introduce a set of desirable properties which constrain cross-impact models. We classify cross-impact models according to which properties they satisfy and stress them on three different asset classes to evaluate goodness-of-fit. We find that two models are robust across markets, but only one satisfies all desirable properties and is appropriate for applications.

Keywords

Cite

@article{arxiv.2004.01624,
  title  = {How to build a cross-impact model from first principles: Theoretical requirements and empirical results},
  author = {Mehdi Tomas and Iacopo Mastromatteo and Michael Benzaquen},
  journal= {arXiv preprint arXiv:2004.01624},
  year   = {2022}
}