BEGIN:VCALENDAR
PRODID:-//eluceo/ical//2.0/EN
VERSION:2.0
CALSCALE:GREGORIAN
BEGIN:VEVENT
UID:8ed22bfa0ecc312ded765227286b936d
DTSTAMP:20260502T221146Z
SUMMARY:Yufei Zhang: An offline learning approach to propagator models
DESCRIPTION:Title: An offline learning approach to propagator models\nAbstr
 act: We consider an offline learning problem for an agent who first estima
 tes an unknown price impact kernel from a static dataset\, and then design
 s strategies to liquidate a risky asset while creating transient price imp
 act. We propose a novel approach for a nonparametric estimation of the pro
 pagator from a dataset containing correlated price trajectories\, trading 
 signals and metaorders. We quantify the accuracy of the estimated propagat
 or 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 strateg
 y purely based on the estimated propagator will encounter suboptimality du
 e to spurious correlation between the trading strategy and the estimator. 
 By adopting an offline reinforcement learning approach\, we introduce a pe
 ssimistic loss functional taking the uncertainty of the estimated propagat
 or into account\, with an optimiser which eliminates the spurious correlat
 ion\, and derive an asymptotically optimal bound on the execution costs ev
 en without precise information on the true propagator. Numerical experimen
 ts are included to demonstrate the effectiveness of the proposed propagato
 r estimator and the pessimistic trading strategy. Based on joint work with
  Eyal Neuman and Wolfgang Stockinger (Imperial College London).
URL:https://www.imperial.ac.uk/events/167880/yufei-zhang-an-offline-learnin
 g-approach-to-propagator-models/
DTSTART;TZID=Europe/London:20231024T140000
DTEND;TZID=Europe/London:20231024T150000
LOCATION:140\, Huxley Building\, South Kensington Campus\, Imperial College
  London\, London\, SW7 2AZ\, United Kingdom
END:VEVENT
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
DTSTART:20231024T140000
TZNAME:BST
TZOFFSETTO:+0100
TZOFFSETFROM:+0100
END:DAYLIGHT
END:VTIMEZONE
END:VCALENDAR