model room
A benchmark for financial modeling that grades the workbook, not the description of it.
Modeling is where analyst work stops being prose. A model can write a fluent paragraph about an LBO and still produce a workbook that does not foot, points at the wrong cell, and breaks the moment an assumption moves. Public evaluations graded the description, not the file.
103 modeling tasks authored and peer-reviewed by practitioners, each shipped with a gold-standard workbook and a grading rubric, across seven categories from LBO and DCF through comps spreads and dataroom summaries. Submissions are graded by recalculating every formula in a real spreadsheet engine, in two modes: fill a provided skeleton, or build from nothing.
Failure became legible. Numerical accuracy, formula correctness and presentation are scored apart from one another, which is how a team learns that a model can wire a workbook correctly and still get the numbers wrong. The categories built on long chains of linked calculations sit lowest every time — something no single headline score would ever show.