Justin Sirignano¶
Affiliation¶
Mathematical Institute, University of Oxford (current). Previously: University of Illinois at Urbana-Champaign (Industrial & Enterprise Systems Engineering) and Imperial College London.
Key contributions¶
- Universal price formation via deep learning — co-author on the canonical paper establishing that a single deep network trained on pooled multi-stock LOB data outperforms stock-specific models (universal-price-formation-sirignano-cont).
- Deep Learning for Limit Order Books (arXiv:1601.01987, 2016) — earlier work proposing deep neural architectures for LOB modelling; precursor to the 2018 universal-features paper.
- Deep BSDE methods and PDE solvers — broader research on deep learning for high-dimensional PDEs (Deep Galerkin Method and related work).
- Deep Learning for Mortgage Risk — applied deep learning to credit / prepayment modelling, demonstrating the Big-Data-beats-small-models thesis outside of HFT.
Notable papers (in this wiki)¶
- universal-price-formation-sirignano-cont — "Universal Features of Price Formation in Financial Markets" (with Cont, 2018).
Connections¶
- rama-cont — collaborator on the universal-features paper.
- universal-price-formation — Sirignano-Cont is the defining empirical reference.
- deep-learning-meets-market-microstructure — Sirignano's work is a foundational contribution to this connection.