Seminar: Eyal Neuman: Offline/Online Learning Approaches to Price Impact Models

PLEASE NOTE THAT THIS IN-PERSON EVENT WILL START ON WEDNESDAY, 25TH OCTOBER, 2023, AT 6:30 PM BST (LONDON TIME) (1.30 PM EDT (NEW YORK TIME)) at G-Research offices.

This event is sponsored by G-Research (Silver Sponsor), First Derivatives plc (Bronze Sponsor) and KX, Inc. (Bronze Sponsor).

This event is hosted by G-Research, Europe's leading quantitative finance research firm: We hire the brightest minds in the world to tackle some of the biggest questions in finance. We pair this expertise with machine learning, big data, and some of the most advanced technology available to predict movements in financial markets.

Venue: G-Research, Whittington House, 19-30 Alfred Place, London, WC1E 7EA

Arrival time of 18:00 (London) and talk starts at 18:30 (London).

Seminar will last between 1 to 1.5 hours, followed by networking / food / drinks.

Please note that your Thalesians Meetup profile must include your full name (and the full names of all your guests) in order to be admitted by the venue, G-Research (health and safety regulations). If it doesn't include it, please email it to alex@thalesians.com along with your profile name.

FULL TITLE: Offline and Online Learning Approaches to Price Impact Models

ABSTRACT

We consider an offline learning problem for an agent who first estimates an unknown price impact kernel from a static dataset, and then designs strategies to liquidate a risky asset while creating transient price impact. We propose a novel approach for a nonparametric estimation of the propagator from a dataset containing correlated price trajectories, trading signals and metaorders. We quantify the accuracy of the estimated propagator using a metric which depends explicitly on the dataset. We show that a trader who tries to minimise her execution costs by using a greedy strategy purely based on the estimated propagator will encounter suboptimality due to so-called spurious correlation between the trading strategy and the estimator and due to intrinsic uncertainty resulting from a biased cost functional. By adopting an offline reinforcement learning approach, we introduce a pessimistic loss functional taking the uncertainty of the estimated propagator into account, with an optimiser which eliminates the spurious correlation, and derive an asymptotically optimal bound on the execution costs even without precise information on the true propagator. Numerical experiments are included to demonstrate the effectiveness of the proposed propagator estimator and the pessimistic trading strategy.
As complementary to the offline learning approach, we also present an online trading algorithm for estimation of the price impact kernel, which alternates between exploration and exploitation phases and achieves sublinear regrets with high probability.
The talk is based on a joint work with Wolfgang Stockinger and Yufei Zhang.

BIOGRAPHY

Eyal Neuman is a Reader (Associate Professor) in mathematical finance and a member of the stochastic analysis research group at the Department of Mathematics, Imperial College London. His research interests are in the areas of probability, stochastic processes and mathematical finance. He is mainly working on interacting particle systems, stochastic partial differential equations and market microstructure. Previously he was a Visiting Assistant Professor at the University of Rochester, NY. Eyal received his PhD in stochastic processes from the Technion.