Accuracy
How we keep the math right
Waterfalls is validated against an independent finance firm's waterfall model across 30 exit values, backed by 303 automated tests, and built on decimal-precision math; every result is shown step by step, so the number you put in a board deck is one you can check.
No credit card required
Exit values matched against an independent finance firm's model, within tolerance.
Automated tests validating every calculation path.
Decimal-precision math, no floating-point drift.
Remainder on every calculation. Total distributed always equals the exit value.
Validated against an independent model
Waterfalls matches an independent finance firm's waterfall model across 30 exit values, and every value matches within tolerance.
A waterfall is only trustworthy if a second, independently built model lands on the same dollars. So we tested ours against one: a finance firm's own model, run across a wide range of deal sizes. Thirty exit values, from small outcomes to large ones (not a single cherry-picked example) and every result reconciled within tolerance. That is external validation, not a self-assessment. We do not claim "audited," "certified," or "guaranteed correct"; we claim the engine matches an independent model across 30 exit values, which is exactly what we tested.
303 automated tests, run continuously
303 automated tests validate every calculation path.
The engine is covered by 303 automated tests spanning liquidation preferences, participation, convertibles, options, carve-outs, precision, and edge cases. They run continuously: every change re-proves the entire engine before anything ships. A spreadsheet has zero tests: a formula can break silently the moment someone drags a cell, and nothing catches it. Waterfalls catches it, automatically, every time.
Decimal-precision math, to the cent
Money is computed to the cent with decimal-precision math (Decimal.js), no floating-point drift.
Money is computed with decimal-precision arithmetic, to the cent, while rates, shares, and prices keep full precision. Ordinary spreadsheet and JavaScript arithmetic can introduce small rounding errors that compound across a multi-tier waterfall: a fraction of a cent here, a fraction there, until the totals no longer tie out. Waterfalls avoids that entire class of error by design, so the dollars at the bottom of the waterfall add back up to the dollars at the top.
Mathematical invariants, enforced on every calculation
Every result is checked: total distributed = exit value · no negative payouts · remainder = 0.
On every calculation, the engine verifies three things: that the total distributed equals the exit value, that no stakeholder receives a negative payout, and that the remainder reconciles to zero, with nothing left unaccounted for. These are not checks you have to remember to run. They run automatically, on every model, and a result that violates them cannot be presented as valid. That is a structural guarantee a spreadsheet fundamentally cannot offer.
Auditable: see every step, defend every number
Every result includes a step-by-step breakdown you can trace and defend.
Every model includes a step-by-step breakdown that traces each tier of the waterfall (liquidation preferences first, then participation and conversion, then the pro-rata remainder) so you can follow exactly how each dollar flowed. A CFO can reproduce it, a lawyer can defend it, and a VC can sanity-check a founder's model line by line. This is the calculation-steps viewer built into every Waterfalls model. For a worked example of the mechanics, see our guide to waterfall analysis and how to calculate it.
Calculation steps: example trace
Secure and private by architecture
Putting a cap table into a tool is a trust decision, so here is the stack, described honestly, as an architecture, not a certification.
- Authentication. Sign-in and identity run on Clerk.
- Data isolation. Your data is stored in Supabase with row-level security, so it is isolated per user at the database layer, not just hidden in the interface.
- Billing and compliance. Polar acts as Merchant of Record, handling VAT and sales-tax compliance and invoicing globally, so payment data and tax handling are managed by a dedicated provider rather than by us.
Built for the terms that break spreadsheets
Accuracy is hard because the terms are hard. Participating preferred with caps, stacked seniority, pari-passu classes, pre- and post-money SAFEs, convertible notes with real interest accrual, underwater options, and management carve-outs are exactly where spreadsheets break quietly, and exactly what the validated engine resolves, in the right order, every time. Each liquidation preference is applied by seniority, the greater-of conversion decision is tested per class, and the remainder is reconciled to zero. The hard cases are the point.
Accuracy, in plain answers
How accurate is Waterfalls?
Waterfalls is validated against an independent finance firm's waterfall model across 30 exit values, and every value matches within tolerance. The engine is also covered by 303 automated tests and computes money to the cent with decimal-precision math, so there is no floating-point drift. It is a modeling tool, not a system of record or a 409A provider, but the math behind each model is built to be checkable.
Can I see how the result was calculated?
Yes. Every result includes a step-by-step breakdown that traces each tier of the waterfall (liquidation preferences, participation and conversion, then the pro-rata remainder) so you can follow exactly how each dollar flowed and defend the number to a board, an LP, or opposing counsel.
Is my data private and secure?
Sign-in and identity run on Clerk. Your data is stored in Supabase with row-level security, so it is isolated per user at the database layer, not just hidden in the interface. Billing runs through Polar as Merchant of Record, which handles VAT and sales-tax compliance globally, so payment data is managed by a dedicated provider rather than by us.
Do I need to import my whole cap table?
No. Waterfalls is a focused modeling tool, not a company-wide onboarding project. You enter only the share classes, instruments, and exit assumptions you want to model, and the engine runs the full distribution from there. You can model a single exit without migrating your entire equity administration.
Run the number you can defend
Externally validated, 303 automated tests, decimal-precision math, and a step-by-step trace on every result. Start modeling in minutes. Every plan runs the same validated engine.
No credit card required
Same validated engine on every plan, see pricing.