In-house trainingFinancial Modelling
Key Outcomes
Structure a model so others can follow it
E.g., separate Inputs, Workings, Outputs and Checks, with consistent formulas across each row and no embedded constants.
Build logic that goes beyond cell formulas
E.g., a three-year forecast that links price, volume and headcount to profit and cash, with a minimum-wage step-up to RM1,700 as a staff cost scenario.
Stress-test with scenarios and sensitivities
E.g., "What if sales fall 15% and customers pay in 75 days instead of 60?", with a base, upside and downside switch.
Check and document the model
E.g., a check cell that flags if cash does not tie out, a peer review for hard-coded numbers, and a change log.
Why teams need this
Many models grow into tangled workbooks with hard-coded numbers, inconsistent formulas and one person who understands them. Panko's 1998 review of seven field audits (88 spreadsheets), as cited in an arXiv paper, found 94% had errors and the average cell error rate was 5.2%. In JPMorgan's "London Whale" case, a task force report described a VaR model run through Excel spreadsheets completed by manual copy-and-paste. One formula divided by the sum instead of the average, which "likely" muted volatility by a factor of two. The Reinhart–Rogoff paper had a spreadsheet error that excluded certain countries, and the corrected relationship was far less dramatic.