Your client asked if they can retire. Your software said “83%.”

See the real outputs before you buy

Download the actual deliverables the product produces — generated by the software itself, not mocked up. No signup, no email.

Or run it on a client plan of your own: the first report is free, no signup, and runs at the full 10,000-path precision of the paid tier — the same report your clients would receive.

📄 Sample client report (PDF, 4 pages) 📄 Sample report with the tax page (PDF, 5 pages) 📊 Forecast-comparison report (PDF) 📑 Forecast-comparison methodology (PDF, 4 pages)

• Browser ↔ backend parity. The in-browser tax engine and the C simulation backend are checked against a shared reference set of 7,934 tax cases — federal ordinary and conversion income, long-term capital gains, Social Security taxation, IRMAA tiers, ACA subsidies, RMD factors and start ages, and all-jurisdiction state brackets — so a figure shown in the browser is the same figure the engine computes. Assumptions and sources are on the public methodology page.
• Recommended-plan optimizer, checked several ways. The withdrawal-order recommendation is a whole-retirement policy search: a set of candidate withdrawal orders — traditional first, taxable first, bracket- and surcharge-tier fills, and the same fills paired with a Roth conversion for the room left over — are simulated on identical market paths, each scored by the present value of lifetime tax including the tax an heir would owe on any traditional balance left at the end and any spending the portfolio could not fund, and the order with the lowest paired-median cost is recommended. Traditional-first is itself always one of the candidates, so the recommendation can only tie or beat it — it is mathematically bounded from below at zero saving, never negative. It runs under constraint invariants: the after-tax spending need is met exactly and the RMD floor is respected. Recommended glide paths (a separate, asset-allocation optimizer) are then independently re-scored with paired common random numbers and a standard error, so a shape is reported as better only when the difference exceeds sampling noise. Results are deterministic and reproducible from run to run.
• Frozen-engine discipline. The same engine powers the public calculator, the downloadable datasets, and the research studies. The canonical datasets are generated directly from that frozen engine rather than re-entered by hand, and studies are corrected only through a dated changelog on the page itself — never silently rewritten.
• Per-lot HIFO tax accounting. Taxable-account sales draw from specific cost-basis lots, highest-in-first-out, and the tax caused by a Roth conversion is funded from real dollars via a HIFO sale — its true marginal after-tax drag, not a cost-free deduction. Documented under tax-lot accounting.
• Source-typed state treatment. Retirement income is typed by source — pension, Social Security, Roth, ordinary — and taxed under each of the 51 U.S. jurisdictions' own rules for that source, including the eight states that tax some Social Security and the states with qualified-retirement-income exemptions. The full per-state policy set is published as an open dataset.

Thirty-four published studies, each with its method, its assumptions and the data behind it — including whether the 4% rule survives forward-looking forecasts, what a bear market at retirement costs across 46 cohorts, and 510,000 Monte Carlo paths across all 51 U.S. tax jurisdictions.

The constants underneath them are published as open datasets — federal and capital-gains brackets, IRMAA tiers, RMD tables, state retirement tax and ACA thresholds — as CSV and JSON under CC-BY-4.0, so any number in a report can be checked against its source.

The independent second opinion

The fastest way to win a prospect from another advisor is an independent stress test of the plan they already have. Enter their current plan, run it through an engine with no stake in the outcome, and leave behind a branded report their current advisor structurally cannot produce — nobody reviews their own plan. Ten minutes of input; a deliverable that does the follow-up for you.

Where client data lives: nowhere near us

QuantCalc has no accounts and no customer database. To run a simulation, the plan’s inputs — ages, balances, spending, allocation — are sent to our engine and the results come back; no client name, identifier or account number is ever part of that request, and nothing is retained on our side. Saved plans live in your browser. There is no vendor data-hosting questionnaire to complete, because there is no vendor data hosting — and because the product never touches client PII, Regulation S-P does not attach to us at all.

The question every solo vendor should answer: if we vanished tomorrow, every report and plan you have exported remains yours as files you already hold, and the methodology behind every figure is fully published, so any number can be reproduced by hand or in anyone else’s tool. The simulation itself runs on our servers, and we would rather say that plainly than pretend otherwise.

Built with the Marketing Rule in view

Reports are built for individual client and prospect meetings. Under the SEC Marketing Rule a projection is hypothetical performance; the rule treats an interactive analysis tool differently when it describes its criteria, methodology, key assumptions and limitations, explains that results may vary with each use and over time, describes the universe of investments considered, and states that outcomes are hypothetical. Every PDF carries those statements on its methodology page, with the full parameter set for your compliance reviewer. How the rule applies to your firm is a determination for your compliance counsel; we are not it.

What advisors should know about the engine

QuantCalc's simulation engine is the same one we use for the public app and the open-data research piece ("Does the 4% rule survive forward-looking forecasts?"). The methodology is documented on a public page; the data behind the research piece is released under CC0 for reproducibility.

• Withdrawal rules. Fixed-real (the 4% rule), Guyton-Klinger guardrails, and VPW — quantified against each other on identical paths in our published study.
• Couples, not just clients. A spouse's own Social Security and pension join the household at their ages; after a death the plan keeps the larger benefit, continues the spouse's pension at its survivor share and steps spending down to the survivor level — on every tier, in the engine and on the report.
• What-if comparisons. Social Security claiming ages (62 / 67 / 70 and the plan's own, benefit restated through the primary insurance amount) and saved scenarios run side by side against the current plan, each row a full simulation, printed as a page of the client report.
• Returns. Six publicly-published Capital Market Expectations sourced from each firm's public publication or widely-circulated financial press. We do not source from paywalled portals.
• Volatility / correlations. J.P. Morgan published correlations (am.jpmorgan.com, checked 2026-09-10) and the 1926–2025 historical volatility series by default; any published source can be selected for each.
• Inflation. Deterministic CPI (2% default, editable); AR(1), regime-switching, and bootstrap models also exposed.
• Tax modeling. 51-jurisdiction state income tax, IRMAA two-factor scaling, ACA cliff, Roth conversion optimizer.

Return sources, volatility, inflation model, sampling regime and the withdrawal rule that ran are documented on the methodology page of every exported report. The sampling regime and generator details are on the methodology page for anyone who wants them.