How to Pass a 3-Statement Modelling Test in 90 Minutes
10 min read
- The 90-minute 3-statement modelling test is the most common format at banks and advisory firms
- The median score is low. Completing a functional model with accurate linkages puts you ahead of most candidates
- Use beginning-of-period debt balances for interest calculations to avoid circular references entirely
- The balance sheet check is non-negotiable. If assets do not equal liabilities plus equity, the model has failed
Three Test Formats and What Each One Actually Tests
Modelling tests come in three formats, and each tests a different skill:
| Format | Time | What It Tests |
|---|---|---|
| Blank sheet, timed | 60-90 min | Speed, Excel fluency, and decision-making under pressure. Can you build a model from nothing without getting stuck? |
| Template, timed | 60-90 min | Formula accuracy and assumption quality. The structure is given; they want to see whether your formulas are correct and your assumptions are defensible. |
| Take-home | 24-72 hours | Depth of analysis, presentation quality, and scenario thinking. More time means higher expectations for formatting, sensitivity tables, and written commentary. |
The blank-sheet 90-minute test is the most common at bulge brackets and elite boutiques. The rest of this article focuses on that format.
The Seven Steps: How to Structure the 90 Minutes
Time allocation matters as much as technical accuracy. A model that is 80% complete with accurate linkages scores better than a model that is 100% complete with broken formulas.
The Circular Reference Problem and How to Avoid It
The classic trap: interest expense depends on the debt balance, which depends on cash flow, which depends on interest expense. This creates a circular reference that breaks the model.
Some modellers use Excel's iterative calculation setting to resolve circulars. This works but is slower to build, harder to audit, and can mask errors. In a timed test, the beginning-balance approach is faster and safer.
Revenue Projection: Where Most Candidates Lose Marks
The default approach is to apply a flat growth rate to last year's revenue. This is acceptable but will not distinguish a strong candidate from an average one.
The better approach, if the filing provides the data: decompose revenue into volume and price, or into business segments. A company with a services business growing at 8% and an equipment business growing at 3% should not be modelled at a blended 5.5%. The segments have different margins, different working capital profiles, and different growth trajectories.
In a 90-minute test, segment-level modelling is only worth attempting if the company clearly reports segments (most do). If it adds 10 minutes and produces a more defensible model, it is worth the time. If it adds 20 minutes and the segments have similar economics, skip it.
Working Capital: The Section Candidates Rush Through
Working capital changes are the most common source of balance sheet errors. The mechanics:
| Item | Formula | Cash Flow Impact |
|---|---|---|
| Accounts Receivable | Revenue × DSO / 365 | Increase = cash outflow (sold but not collected) |
| Inventory | COGS × DIO / 365 | Increase = cash outflow (bought but not sold) |
| Accounts Payable | COGS × DPO / 365 | Increase = cash inflow (bought but not yet paid) |
The change in each item flows to operating cash flow. Calculate the absolute value from the days ratio, then take the year-over-year change. An increase in AR is a use of cash. An increase in AP is a source of cash. Get the signs wrong and the balance sheet will never balance.
Debt and Capital Allocation: The Advanced Layer
In a basic test, keeping debt flat is acceptable. In a more advanced test, or when the company has stated a capital allocation policy, modelling debt repayment and share buybacks shows sophistication.
The framework: calculate free cash flow after dividends. Determine the minimum cash balance the company needs to operate (2-3% of revenue, or the level stated in the filing). Any excess cash above that minimum can be allocated to debt repayment, share buybacks, or acquisitions based on management guidance.
The Balance Check: Non-Negotiable
Add a row at the bottom of the balance sheet: Total Assets - Total Liabilities - Total Equity. This should equal zero in every forecast year. Conditional format it to turn red if non-zero.
If it does not balance, the most common causes in order of frequency:
- Working capital change signs are inverted on the CFS
- Ending cash on the CFS does not link to cash on the BS
- Retained earnings formula is missing a component (forgot dividends, or double-counted net income)
- PP&E does not tie to capex and D&A
- A debt issuance or repayment on the CFS was not reflected on the BS
What Evaluators Actually Grade On
| Criterion | Weight | What They Check |
|---|---|---|
| Accuracy | ~40% | Do the statements link correctly? Does the balance sheet balance? Are formulas right? |
| Assumptions | ~25% | Are revenue growth, margins, and working capital days reasonable and derived from historicals? |
| Structure | ~20% | Is the model organised logically? Can someone else navigate it? Consistent formatting? |
| Completeness | ~15% | Did you finish? A complete model with minor errors beats an incomplete one every time. |
Practice Approach
Pick any public company with a simple business model (one or two segments, minimal M&A). Pull 3 years of historicals from the 10-K. Set a 90-minute timer. Build the model. Check the balance sheet. Repeat with a different company until the process is automatic.
After 5-6 practice models, the mechanical steps become muscle memory and the 90 minutes starts to feel comfortable. The candidates who struggle are the ones who practised by reading about modelling rather than building models.
Take Your Preparation Further
Download our free Accounting Cheat Sheet for the complete three-statement linkage reference. For a hands-on model with all linkages built, see the 3-Statement Financial Model.
Ready for personalised feedback? Book a 1-on-1 mentoring session with an experienced IB/PE professional.