Monte Carlo simulation

Rather than assume one fixed rate of return, FEAsible plays your plan forward through thousands of different randomized futures — each a distinct sequence of monthly market returns drawn from your assumptions. Your plan success is simply the share of those futures in which the money lasts.

So a run of 1,000 paths × 420 months × 6 assets means roughly 2.5 million individual asset-return figures were generated and rolled up — that's the 'simulated asset-returns' number on the results page. More paths means a steadier, less luck-dependent estimate.

Return models & stress testing

The randomized futures can be drawn two ways, and the choice is yours. The default parametric model samples from your return and volatility assumptions. The historical models instead use the real record — Aswath Damodaran's 1928–2025 annual returns/CPI series (NYU Stern) — replaying it in sequence, or resampling it in multi-year blocks, so the ordering of good and bad years is one that genuinely occurred rather than one a formula invented.

In every model, returns and inflation are drawn together, not independently — so their real-world relationship is preserved: a high-inflation year tends to arrive with the weaker real returns it historically accompanied (the “stagflation corner”), instead of the two being paired at random.

Because no single model is the whole truth, the stress test runs your plan through several of them at once and reports the spread — the success range and the safe-withdrawal band across models. A plan that only holds under one set of assumptions is then easy to spot.

Antithetic pairs (a variance-reduction trick)

When enabled, the simulation generates random futures in mirrored pairs: for every randomly drawn sequence, it also runs its mirror image (the good-luck draw paired with its bad-luck opposite). Averaging the pair cancels out a lot of plain luck, so you get a more accurate answer from the same number of paths. It is purely a numerical efficiency technique — it does not change your plan or its assumptions.

Seed, engine version & config fingerprint

• Seed: the starting point for the random-number generator. The same seed with the same inputs reproduces the exact same run, every time — results are deterministic, not a different roll of the dice each visit.
• Config fingerprint: a short code (a hash) computed from all of your inputs. If two runs show the same fingerprint, they were run on identical inputs — handy for confirming two scenarios are truly comparable.
• Engine version: the calculations run in a separate, versioned engine (FEAsibleEF); each result records which release produced it, so a number can always be traced to the exact logic that made it — and, with the seed and fingerprint, reproduced.

How the optimizers search

An optimizer doesn't guess — it enumerates and scores concrete strategies. For Roth conversions it builds a menu of natural boundaries each year (the top of a tax bracket, or just under a Medicare premium tier), because the best conversion almost always lands exactly on one of those edges. It then assembles many full multi-year candidate schedules from that menu, tries them under different orders of which account/owner to convert first, projects each one, and ranks them by lifetime tax.

So "evaluated 864 candidate schedules across 3 owner orderings and 25 convertible years" means it scored 864 complete conversion plans — that thoroughness is why the recommendation beats hand-tuning. The same enumerate-and-score approach chooses your withdrawal order and Social Security claim age. The winner is what you lock in as a strategy layer.

Sustainable-spending solver

The 'sustainable spending' figure is the highest annual spending that still holds your success target (90%). FEAsible finds it by bisection: guess a spending level, run the simulation, then halve the search range up or down based on whether it passed — repeating until it converges. 'Solved in 5 passes' means five such guess-and-check rounds pinned it down.

How the spending rules differ

The variable spending rules on the Expenses page are not five flavours of the same idea — each one bounds a different dimension of your spending, and each answers a different fear. Every rule governs your portfolio withdrawal; Social Security and pensions are spent on top, and the floor/ceiling you set bound your total spending whatever rule runs. Picking a rule starts it at its author's published numbers; every dial stays yours to move afterward.

• % of portfolio — decides the LEVEL, from current wealth, every year: withdraw a fixed share of whatever the portfolio is worth. Fully market-linked, and mathematically it can never run the portfolio to zero — the trade is that your spending swings as much as your balance does. For the person whose first fear is running out.
• Bengen floor & ceiling (Bengen, 2001) — the same market-linked withdrawal, but never below 90% nor above 125% of your first year's withdrawal in today's dollars. Market-linked with the swings capped, from the author of the original 4%-rule research.
• Clyatt 95% (Clyatt, “Work Less, Live More”, 2005) — decides the RATE OF CHANGE: withdraw 4% of the current portfolio, but never less than 95% of last year's withdrawal. Rises track the market; falls arrive as a slow squeeze instead of a shock. For the person a sudden 30% pay cut would wreck.
• Guyton-Klinger (Guyton & Klinger, 2006) — decides the TRIGGER: a steady, inflation-adjusted draw that steps down or up 10% only when your withdrawal rate drifts 20% off course — plus no cuts in the final 15 years, and inflation raises skipped after a down year. Steady, with rare corrections. The Simplified guardrails preset is the decision bands alone — the rule this app ran before 1.2, kept under its honest name.
• Kitces ratchet (Kitces, 2015) — decides the UPSIDE ONLY: start conservatively at 4%, and draw 10% more each time the portfolio grows 50% past its retirement value (at most every 3 years). It never cuts. For the person who'd lock in a number that's too low and under-live for 30 years.
• VPW (Bogleheads) — decides the HORIZON: withdraw a rising share as you age, computed like a mortgage payment over your remaining years, so the portfolio draws down smoothly to your planning end instead of running dry or leaving a windfall. Declining late-year draws are the design. For the person who doesn't want to die with money they could have enjoyed.

Every preset's constants are pinned to its source by tests, and the rule you build is named honestly: change a defining lever and it relabels itself “Custom (based on …)”. The preview on the Expenses page shows how your exact lever set responds to an illustrative crash and boom.