Please note javascript is required for full website functionality.

Our Key Concepts

CRaFT

CRaFT is SumProduct’s proprietary financial modelling ideology. Rather than adopt a 300-page document of Standards, we believe a good financial model should exhibit the following four key qualities:

  • Consistency
  • Robustness
  • Flexibility
  • Transparency

If a model meets these four key qualities, it should adhere to any professional standards sought. These four ideas are explored further below.

Consistency

  • Ensures fewer mistakes, promotes model integrity and makes it easier for developers and users alike to understand models more quickly
  • Assists the model developer by building structured models efficiently and effectively:
    • incorporating pre-defined formats and styles in a coherent manner
    • requiring formulae to be copied uniformly across ranges
    • presenting periodic data and formulae in columns consistently, thereby ensuring similar structures for similar worksheets

Robustness

  • Models must be materially free from error, mathematically accurate and readily auditable
  • Develop / review spreadsheets under a philosophy of keeping things simple, consistent and transparent
  • Hence, models should provide strong, inherent levels of structural, logical and formula sturdiness
  • Ideally, well-built models should contain:
    • no hidden macros
    • in-built error, sensitivity and alert checks
    • audited, established formulae

Flexibility

  • Models should have the structure to provide invaluable “what if?” scenario / sensitivity analysis
  • End users must be able to change key assumptions within agreed parameters

Transparency

  • Many modellers often forget that key decision makers base their choices on printed material
  • Models need to be clear, concise and cohesive
  • Modellers should create assumptions, calculations and outputs that are recognisable instantly and user-friendly
  • For example, the developer / reviewer should consider:
    • the logical flow and dynamic linking of the model
    • the number of categories required for particular calculations
    • the nature of the assumptions together with their assumption entry methods at the design stage

Newsletter