PhD Qualifying-Exam

ADAPTING MODELS FOR QUANTITATIVEFINANCE: A SURVEY OF POST-TRAINING

The Hong Kong University of Science and Technology (Guangzhou)

Data Science and Analytics Thrust

PhD Qualifying Examination

By Ms. LIU, Jiaqi

Abstract

Financial models need to adapt under many scenarios for example when they move across
markets or when market regimes change. Each setting can bring different data and economic objectives. As quantitative finance relies more on large models, training a new
model from scratch for each change can be costly. Post-training addresses this problem
by adapting what a model has already learned. This survey examines post-training for
financial prediction and decision-making. We organize the literature by the purpose of
each adaptation stage, covering cross-market transfer, adaptation of numerical and language representations, policy learning under economic objectives, and continual adaptation. Source representations help when they retain information the target task needs. An
economic objective can improve trading decisions even when forecast error rises. When
feedback is delayed, new observations can refresh forecast inputs or guide the selection
of historical examples. The model can then learn from outcomes available at each update.
We discuss what to reuse, how to account for risk and cost in training, and when to update
models as markets change.


PQE Committee

Chair: Prof. TANG, Jing (onsite)

Prime Supervisor: Prof. LI, Jia (onsite)

Co-Supervisor: Prof. GUO, Jian (online)

Examiner: Prof. ZHANG, Yongqi (onsite)

Date

04 December 2026

Time

15:30:00 - 16:30:00

Location

E3-105, HKUST (GZ)

Event Organizer

Data Science and Analytics Thrust

Email

dsarpg@hkust-gz.edu.cn