Credit scoring refers to the collection of quantitative techniques used to assess the risk of lending to consumers, and it stands as one of the most successful applications of statistical and operations research modeling in modern finance. At its core, the objective is to assign a probability of default to a loan applicant. As Thomas explains, this probability is not arbitrary; it depends on a relatively large number of variables that determine an individual's ability to repay debt.
Highly interpretable; standard industry benchmark; mathematically robust.
The book also addresses the critical area of Profit Scoring. While traditional models focus on the probability of default, profit scoring shifts the lens to the overall value a customer brings to the firm. This involves balancing the interest income and fees against the costs of capital and potential losses. By focusing on profitability, lenders can optimize their portfolios to maximize returns rather than just minimizing risk. credit scoring and its applications by l c thomas hot
Prone to overfitting if tree depth is not strictly constrained. (Modern Evolution)
Sorting and assessing raw data to ensure it is reliable ("Data Massaging"). Credit scoring refers to the collection of quantitative
The book distinguishes between different types of models based on their scope and application:
: Later editions and related works by Thomas incorporate Markov chains and survival analysis to model repayment behaviors over time. This involves balancing the interest income and fees
“Given a delinquent customer, what action will yield the highest recovery?”
: This phase determines whether to extend credit to a new applicant. It relies on data provided at the point of application paired with credit bureau records.