Khraishi, Raad; (2025) Machine Learning for Financial Decision Making: Pricing, Customer Lifetime Value, and Promotional Strategies. Doctoral thesis (Ph.D), UCL (University College London).

Text PhD_Thesis Submitted.pdf - Accepted Version Access restricted to UCL open access staff until 1 June 2026. Download (3MB)

Abstract

Financial institutions increasingly rely on data-driven techniques to navigate the complexities of consumer finance, including pricing, marketing, and customer relationship management. This work presents four interrelated research contributions that harness modern machine learning (ML), reinforcement learning (RL), and agent-based modelling (ABM) to address these challenges. First, we propose an offline RL framework for setting prices for a single debt product. Our method, validated on real and synthetic data, learns directly from historical datasets, avoiding risks of live experimentation and outperforming an industry-standard benchmark by around 20% in expected profit with only limited changes to historical pricing. Next, we develop an ML-based approach for estimating customer lifetime value (CLV) across multiple products, focusing on both acquisition and profitability, which has been implemented at a large UK bank. We then construct an ABM of the banking market, calibrated to historical data, to analyse competitive and consumer impacts of different credit card promotion strategies, enabling controlled testing of new marketing and pricing policies and offering practical insights while minimising real-world risks. Finally, we integrate these threads with a multi-product, model-based, offline RL framework that jointly optimises prices to maximise CLV. Empirical results using a synthetic ABM demonstrate that our approach outperforms standard single-product benchmarks by approximately 30% in discounted return over 12- and 24-month horizons. Overall, this thesis underscores the potential of ML and ABM to enhance financial strategies in commercial banking, enabling optimised pricing, enhanced relationship management, safer experimentation, and improved profitability.

Download activity - last month

Export as JSONXMLCSV

Download activity - last 12 months

Export as CSVXMLJSON

Downloads by country - last 12 months

Export as XMLCSVJSON

Archive Staff Only

View Item