Which statement is essential when building a scorecard?

• A. Review data from past borrowers to determine which characteristics are most useful
• B. Historical data must be of sufficient size to make analysis valid
• C. Need to review characteristics for both good and bad clients
• D. Small business loans must look at both business and PG data

Answer: (B)

Building a scorecard relies on having enough historical data to produce reliable, stable estimates of how borrower characteristics relate to outcomes. When the data set is large enough, the model can distinguish true patterns from random noise, giving weights that generalize to new applicants. If the data are too sparse, estimates become volatile, the model overfits the sample, and the resulting scores can misclassify risk.

While identifying which characteristics are useful, reviewing both favorable and unfavorable borrowers, and deciding which data domains to include (such as business versus personal data) are all important practical steps, they depend on having sufficient data to yield trustworthy results. Without adequate data size, even well-chosen features and thoughtful data scope won’t produce a valid scorecard.