Stress-Testing Your Valuation Assumptions

By Nazli Hangeldiyeva

A valuation model is a collection of assumptions connected by arithmetic. The intrinsic value it produces is conditional: if revenue grows at 8%, and margins expand to 18%, and capex runs at 5% of revenue, and the discount rate is 9%, then the stock is worth $74 per share. Change any of those conditions and the conclusion changes with it. Sensitivity analysis is the process of systematically varying those assumptions to understand which ones matter most, how wide the valuation range is, and at what point the investment thesis breaks.

Every professional investment memo includes sensitivity analysis. At Bridgewater Associates, every investment thesis is stress-tested against multiple macroeconomic scenarios. At Baupost Group, Seth Klarman has described the process of identifying all the ways an investment can lose money before evaluating how it can make money. The purpose is not to achieve false precision but to understand the distribution of possible outcomes and to determine whether the risk-reward profile is attractive at the current price.

Two-Variable Sensitivity Tables

The most common form of sensitivity analysis is the two-variable data table, where the intrinsic value per share is recalculated across a grid of two key assumptions. The standard DCF sensitivity table uses WACC on one axis and terminal growth rate on the other.

A simplified example for a consumer goods company:

If the stock trades at $70, this table shows that the investment offers a margin of safety across most assumption combinations except the most pessimistic (high WACC, low growth). If the stock trades at $90, the investment only makes sense at the most optimistic end of the range.

Additional two-variable tables worth building:

• Revenue growth rate vs. operating margin
• Revenue growth rate vs. WACC
• Exit multiple vs. terminal growth rate
• Year-5 revenue vs. terminal operating margin

Scenario Analysis

While sensitivity tables test continuous variation in two parameters, scenario analysis tests specific, internally consistent sets of assumptions that tell a coherent story about different futures.

Base case. The analyst's best estimate of the most likely outcome. Revenue grows at 7%, margins stabilize at 16%, and the company executes its strategy without major disruptions. This is the scenario that produces the headline intrinsic value.

Bull case. An optimistic but plausible scenario. Revenue growth accelerates to 10% due to new product traction, margins expand to 19% on operating leverage, and the company benefits from a favorable macro environment. This scenario defines the upside potential.

Bear case. A pessimistic but plausible scenario. Revenue growth slows to 3% due to competitive pressure, margins compress to 13% as the company invests defensively, and the macro environment deteriorates. This scenario defines the downside risk.

Stress case. A severe scenario that tests the company's survivability. Revenue declines 10% in a recession, margins collapse to 8%, and the company needs to access debt markets at elevated rates. This scenario is not the most likely outcome, but understanding the consequences is part of responsible analysis.

Each scenario should produce a complete set of projected financial statements and a corresponding intrinsic value. The range from bear to bull case establishes the uncertainty band around the base case estimate.

Identifying the Key Drivers

Not all assumptions are equally important. Sensitivity analysis reveals which variables have the greatest impact on the valuation, allowing the analyst to focus research time on the assumptions that matter most.

Tornado diagrams visualize the relative impact of each assumption by showing how the intrinsic value changes when each variable is moved individually by the same percentage (e.g., +/- 10%). The variables with the widest bars (largest impact) are the key drivers.

For most companies, the top drivers typically include:

1. Revenue growth rate (especially for growth companies)
2. Terminal value assumptions (growth rate or exit multiple)
3. Operating margin (especially for companies with variable margin profiles)
4. Discount rate (WACC)
5. Capital expenditure intensity (for capital-heavy businesses)

Once the key drivers are identified, the analyst should:

• Invest the most research time in understanding these variables
• Gather the most evidence supporting the assumptions
• Monitor these variables most closely as new data becomes available
• Build specific scenarios around different outcomes for these drivers

Monte Carlo Simulation

For analysts comfortable with more advanced techniques, Monte Carlo simulation assigns probability distributions to key assumptions (rather than discrete point estimates) and runs thousands of iterations, each randomly sampling from those distributions. The result is a probability distribution of intrinsic values rather than a single number or a simple range.

For example, instead of assuming revenue growth is exactly 8%, the analyst specifies that revenue growth follows a normal distribution with a mean of 8% and a standard deviation of 3%. The simulation randomly draws growth rates from this distribution thousands of times, producing a distribution of implied values.

The output might show:

• 10th percentile value: $52 per share
• 25th percentile: $62
• Median (50th percentile): $74
• 75th percentile: $87
• 90th percentile: $105

This probabilistic framing is more intellectually honest than a single point estimate because it acknowledges the uncertainty inherent in all projections. If the stock trades at $65 and the median Monte Carlo output is $74 with a 75th percentile of $87, the risk-reward profile is favorable. If the stock trades at $95 and only 10% of simulations support that price, the margin of safety is thin.

Monte Carlo simulations can be built in Excel (with add-ins like @Risk or Crystal Ball) or in Python, R, or other programming languages. The key input is not the simulation tool but the analyst's judgment about the distributions of key assumptions: their mean, dispersion, and any correlations between them (e.g., revenue growth and margin may be positively correlated if higher revenue drives operating leverage).

Reverse Engineering: What the Market Is Pricing

One of the most valuable applications of sensitivity analysis is reverse engineering the assumptions implied by the current stock price. Rather than starting with assumptions and deriving a value, start with the market price and determine what assumptions are required to justify it.

If a stock trades at $120 per share, what combination of revenue growth, margin, and discount rate produces a DCF value of $120? If the answer requires 12% annual revenue growth for a decade with 25% operating margins, the analyst can then assess whether those assumptions are realistic. If they are aggressive, the stock may be overvalued even if the company is excellent. If they are conservative, the stock may be undervalued.

This technique is particularly useful for companies trading at elevated multiples. When Nvidia traded at 60x forward earnings in 2024, the question was not whether Nvidia was a great company (it clearly was) but whether the stock price already reflected the enormous growth expectations. Reverse-engineering the implied assumptions and comparing them to the analyst's own projections reveals whether the market's optimism is justified.

Margin of Safety in Practice

Benjamin Graham introduced the concept of margin of safety: buying at a price sufficiently below intrinsic value that the investment remains profitable even if assumptions prove somewhat wrong. Sensitivity analysis quantifies the margin of safety by showing the range of outcomes across different assumptions.

A stock trading at $50 with a base-case intrinsic value of $75 has a 33% margin of safety on the base case. But if the bear-case value is $40, the margin of safety disappears in a pessimistic scenario. If the bear-case value is $60, the margin of safety exists even under adverse conditions.

The required margin of safety depends on the uncertainty of the analysis. For a regulated utility with predictable cash flows, a 15-20% discount to intrinsic value may be sufficient. For a cyclical industrial company with volatile earnings, a 30-40% discount is more appropriate. For a biotech company with binary clinical trial outcomes, even a 50% discount may not provide adequate protection.

Practical Stress-Testing Checklist

Before finalizing a valuation, run through the following stress tests:

Revenue stress. What if revenue growth is half the base case assumption? Is the company still solvent? Does the stock still offer upside?

Margin stress. What if operating margins compress by 300-500 basis points due to competitive pressure, input costs, or pricing power erosion? How does free cash flow change?

Rate stress. What if the discount rate increases by 150-200 basis points due to a shift in the risk-free rate or a widening equity risk premium? This is not hypothetical; it happened in 2022 when the 10-year Treasury yield moved from 1.5% to 4.3%.

Capital requirement stress. What if capex needs to increase by 50% to maintain competitive position? This is relevant for companies facing technology transitions or capacity constraints.

Duration stress. What if the terminal value growth rate is 1% instead of 3%? This tests sensitivity to the most uncertain and most impactful assumption.

Liquidity stress. What if the company cannot refinance debt at maturity? Check the debt maturity schedule and assess the company's access to credit markets.

No model can predict the future. But a model that has been thoroughly stress-tested provides an honest assessment of the range of outcomes and the conditions under which the investment thesis fails. That honesty is worth more than any false sense of precision.

Written by

Nazli Hangeldiyeva

Co-Founder of Grid Oasis. Political Science & International Relations, Istanbul Medipol University.

View full profile →