Most market-sizing work ends in a single number. The PowerPoint says "TAM is $2.3 billion" and the conversation moves on. What that number conceals is that nearly every input — population eligibility, per-member fees, EBITDA multiples, growth rates — is itself uncertain. The pretense of precision actively misleads.
For a private equity team evaluating a healthcare services acquisition, this wasn't acceptable. The diligence question wasn't just "what is the value?" — it was "how confident can we be in that value?" If the realistic range spans 2x or 3x, the deal looks very different than if the range is tight.
Replacing Point Estimates with Distributions
We replaced the standard point-estimate TAM with a Monte Carlo simulation. Each input — eligible lives by state, per-member fee, addressable share, EBITDA multiple — was specified not as a single value but as a probability distribution: a most-likely value flanked by a plausible range. The simulation drew thousands of scenarios from those distributions and compounded the inputs through to a final value, producing a full picture of what the business could be worth, not just one estimate.
A point estimate says "this is the answer." A distribution says "here's the answer, here's how much I'd believe it, and here's where the risk lives." For a deal team, the second answer is far more useful.
What the Simulation Showed
seven targeted states
the simulation
high outcomes
The simulation suggested a most-likely value of $231 million for the seven targeted states, but framed within a plausible range of $173 to $377 million. Two things mattered about that picture beyond the headline number. First, the spread itself — over a 2x ratio between low and high — quantified the diligence risk in a way no point estimate could. Second, the distribution was asymmetric: the upside tail was longer than the downside, driven by EBITDA-multiple uncertainty.
Why This Mattered for the Deal
The team didn't have to commit to one number. They could underwrite to the most-likely scenario but stress-test the deal against the 10th-percentile outcome — the question becoming "does this still work if we land closer to $173 million?" rather than "is $231 million correct?" That's a sturdier basis for going forward, walking away, or repricing.
The same framework applies anywhere uncertainty compounds across multiple inputs — new-market entry decisions, infrastructure-investment cases, capacity planning, demand forecasting in volatile categories. Whenever inputs aren't precisely known and the conclusion depends on stacking several of them together, simulating the result honestly will outperform a single deterministic estimate. The harder part is usually political: getting people comfortable presenting an answer that admits its own uncertainty. The math is the easy part.
Working on a diligence, sizing, or planning question where the inputs are genuinely uncertain? Say hello.