There is a particular kind of failure that does not feel like failure while it is happening. The analysis is rigorous. The model is well built. The assumptions are defensible. The sensitivity tables are thorough. Every number checks. And the decision the analysis produces is wrong.
This happens when the spreadsheet is solving the right equation at the wrong resolution.
Resolution is an analytical variable
Every analysis operates at a level of resolution. A country-level analysis resolves the country. A deal-level analysis resolves the deal. The two are not interchangeable, and the most consequential error in Brazilian investment analysis is applying country-level resolution to a deal-level decision.
The error is invisible from inside the model. The model does not know its own resolution. It processes the inputs it is given and produces outputs at the resolution of those inputs. If the inputs describe the country, the outputs describe the country — and the decision the model recommends is a decision about the country, dressed up as a decision about the deal.
You can model a Brazilian real estate investment with a discount rate calibrated to the second decimal, a vacancy assumption drawn from national data, and an exit cap rate benchmarked against aggregate market figures — and produce a number that is precise, defensible, and irrelevant to whether this specific property in this specific municipality can be activated on the timeline you assumed.
The three resolution errors
Aggregate inputs for specific decisions
The model uses national or sector-level figures — average licensing timelines, market cap rates, typical activation costs — for a decision that depends on the specific values in the specific jurisdiction. The averages are real. They are also not the numbers that will determine this deal.
Smoothed time for lumpy reality
The model assumes a timeline that flows smoothly from acquisition to revenue. The operational reality is lumpy: a licensing process either completes or it does not, on a specific date, with specific consequences. A six-month delay is not a smooth adjustment to the model. It is a discontinuity the model never represented.
Expected case as base case
The model treats the expected case as the planning case, with the conservative case relegated to a sensitivity table. In Brazilian operational reality, the conservative case is frequently the realistic case — and a model that treats it as an edge scenario is mispricing the central risk.
Why precision masks the error
The danger of a well-built model is that its precision is mistaken for accuracy. A model that produces a number to the second decimal feels rigorous. The rigor is real at the level of arithmetic. It is absent at the level of inputs.
Precision in computation does not compensate for error in resolution. A model that precisely processes country-level inputs produces a precisely wrong answer to a deal-level question. The decimal places create confidence that the resolution does not justify.
This is why experienced investors who would never accept a sloppy model can still be led to a wrong decision by a clean one. The cleanliness is not the issue. The resolution is.
Building at the right resolution
The correction is not a better model. It is better inputs — inputs drawn from the operational layer of the specific deal rather than from the aggregate layer of the country.
The activation cost estimated from comparable projects in the same municipality, not from national averages. The timeline built from the specific licensing authority's track record, not from sector-typical figures. The vacancy and rent assumptions sourced from the specific submarket through local professionals, not from public indices. The exit assumptions modeled against the actual liquidity of the specific market, not against aggregate benchmarks.
With deal-level inputs, the same model becomes a powerful instrument. The arithmetic was never the problem. The resolution of the inputs was. Fix the resolution, and the spreadsheet that was leading you to the wrong decision starts leading you to the right one.
Built a model you trust but can't fully validate the inputs?
A Brazil Complexity Brief pressure-tests the deal-level assumptions — timelines, activation costs, municipal capacity — that determine whether the model is solving the right problem.
Request a Brief — US$490 →Common questions
Can a financial model be wrong even if the math is correct?
Yes. A model with correct arithmetic can produce a wrong decision if its inputs are at the wrong resolution — country-level data applied to a deal-level decision. The math is right; the answer is wrong because the inputs describe something other than the actual deal.
What does analytical resolution mean in investment analysis?
Resolution is the level of specificity at which an analysis operates. Country-level resolution describes aggregates; deal-level resolution describes a specific transaction. Applying the wrong resolution to a decision is a structural error that precision in computation cannot fix.
How do I know if my Brazilian investment model is built at the right resolution?
Check whether the inputs come from the specific deal — the specific municipality's licensing record, the specific submarket's transacted comparables, the specific structure's tax consequences — or from national and sector aggregates. If the inputs are aggregate, the model is at country resolution regardless of how detailed it appears.