What it does. It simulates an irrevocable trust that pays one annual distribution to all living descendants, with each person's share weighted by generation or age, and shows how the trust and each generation's payouts could turn out across thousands of simulated families and market histories.

How it works. Each simulation builds a random family (births, deaths and the number of descendants by generation, with optional trends in family size and life expectancy for people born later) and runs the trust year by year. Investment returns are drawn at random each year (Monte Carlo), generated by a layered model whose returns change over time (Layered), held constant (Simple), or taken from real US history by picking a random start year and running forward in order (Historical). Payouts follow the rule you choose. Taxes are either a flat drag or a detailed estimate of trust-level tax on retained dividends and on gains from shares sold, with an optional beneficiary-level income tax. Percentiles show the 10th, 50th (median) and 90th percentile outcomes across simulations. All dollars are in today's dollars unless you switch to nominal in the single-family view.

Layered return model. Each simulation stacks four layers, all in real (after-inflation) terms. (1) Centuries: the simulation draws its own long-run return around the one you enter, optionally drifting by a set amount per century. (2) Decades: the expected return wanders around that long-run return in eras with a set half-life; when prices fall, expected future returns rise, so part of a fall is made up later (the “recovery strength” setting, following the predictive-system approach of Pástor and Stambaugh). (3) Years: markets switch between normal years and crisis years (about one year in five; crisis spells average about three years) with higher volatility and lower returns, plus optional slow shifts in volatility over decades. (4) Rare events: disasters cause permanent losses that are not made up afterward. Inflation varies from year to year and rises in crisis years. The long-run return you enter already allows for disaster losses, so the median compounded return stays close to it. With “treat these settings as uncertain” on, each simulation also draws its own era length, swing size, recovery strength and disaster chance, which approximates simulating from a Bayesian posterior. Defaults are starting assumptions; they will be replaced by values fitted to global historical data.

Main limits.

• The long-run average return is the biggest driver of results and cannot be known. Over 100 to 250 years a difference of a point or two changes outcomes by an order of magnitude. Use the conservative and optimistic setups and the uncertain-average-return option.
• In the Monte Carlo model yearly returns are independent of each other, while real markets show longer cycles. The Layered model adds eras, crises and disasters, but its settings are hand-set assumptions, not yet fitted to data.
• The trust is assumed to be fully exempt from the federal generation-skipping transfer (GST) tax, meaning it is funded within the donor's GST exemption ($15M per person in 2026) with the exemption allocated to it. A trust that is partly or wholly non-exempt would owe a 40% GST tax as wealth passes to grandchildren and later generations; that tax is not modeled.
• Taxes use flat rates you enter, not progressive brackets. The trust is assumed to be based in a state that does not tax trust income; beneficiaries' state income tax is optional. Spouses are not beneficiaries.
• Fertility and life expectancy can trend steadily over time, but the trends are straight lines you choose, not forecasts, and they apply to everyone in the family alike.
• Some figures (payout medians, lifetime percentiles) are read from compact histograms and can differ from exact values by about 1%. Confidence ranges, when shown, reflect only the number of simulations.
• Defaults for returns, volatility, yields and tax rates are illustrative assumptions, not forecasts.

Sources. Historical annual returns for the S&P 500 (with dividends), 10-year US Treasury bonds and US inflation, 1928 to 2023, are from Aswath Damodaran, NYU Stern, Historical Returns on Stocks, Bonds and Bills. The historical stock data is US-only, so it is likely optimistic for a global stock index.

Privacy. Everything runs in your browser. Your settings are not sent to or stored by any server. A share link puts your settings in the web address after the # sign, which browsers do not send to the site. The page loads the Chart.js charting library from the cdnjs content network, so your browser contacts that network when the page opens.

Terms of use. This calculator is for general information and education only. It is not financial, investment, tax or legal advice, and using it does not create an advisory, attorney-client or other professional relationship. Results are hypothetical simulations based on assumptions that may not hold, and past performance does not predict future results. The tool is provided "as is" without warranties of any kind. To the fullest extent permitted by law, the author is not liable for any loss or decision arising from its use. Consult qualified professionals before acting.

Copyright and license. © 2026 Meeks. The code is released under the MIT License, reproduced below. The Chart.js library, loaded from the cdnjs network, is licensed under the MIT License by its authors. Historical market data is credited to its source above.

MIT License

Copyright (c) 2026 Meeks

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

Version. October 2026.