Retirement & Portfolio Planning Calculators

For the question most people arrive with — what withdrawal rate did this allocation survive historically — the free tools are not a compromised version of anything, and you should start there. What money buys is everything that is not about markets: tax treatment by account type, Roth conversions, Social Security timing, health insurance before 65, and a plan you can save rather than rebuild from a URL. Before you read any output, find out which of three methods produced it, because the number means something different in each.

• Historical cycles, free, no account — cFIREsim or FI Calc.
• Accounts, taxes and conversions modelled properly — Boldin at $144 a year or ProjectionLab at $129.
• What an allocation actually did, in your own currency — Portfolio Charts.
• Backtesting a portfolio of real funds — Portfolio Visualizer, or PortfoliosLab at $10 a month, which uses only the funds' own daily prices and optimises the weights seven ways.
• European UCITS funds, in euros — Backtest by Curvo, free, on index data back to 1970.
• The longest daily history, simulated where the funds did not exist — testfolio, to 1885, with full history on the free tier.

The paid planners also differ in what kind of answer they return. Boldin, WealthTrace and ProjectionLab give a probability that a plan survives; MaxiFi Planner solves for the spending level you can hold for life instead; and Open Social Security answers one decision, when to claim, for free. Between the free historical simulators and those planners sits FIREproof, from cFIREsim's author: the same historical cycles, with the account types and US tax rules cFIREsim leaves out, at $10 a month.

The simulation method is an assumption, not a feature

Every tool here turns a plan into a single number, and that number means something different depending on how it was produced. Vendors list the method as a bullet point. It is the entire interpretation of the result.

Historical sequences. Replay the plan through every real start year in the dataset — retire in 1929, in 1937, in 1966, in 2000 — and count how many of those runs ended with money left. This is what cFIREsim and FI Calc do, and it is honest about one thing the alternatives are not: every sequence it shows you actually happened, in that order, with that inflation. The limitation is arithmetic. A dataset covering 1871 to today yields a hundred-odd rolling 30-year windows, and consecutive windows share 29 of their 30 years, so the number of genuinely independent observations is somewhere between five and ten. The mid-century record is doing most of the work, and a run of years that never occurred once is simply not in the sample.

Monte Carlo. Draw thousands of random return sequences from a distribution and count the survivors. This escapes the sample-size problem and introduces a worse one: the answer is entirely a property of the distribution, and the distribution is a vendor default you are seldom shown. Whether returns are drawn independently each year or with mean reversion, what the assumed standard deviation is, whether inflation is correlated with anything — these choices move the success rate by more than any input you will type. Independent annual draws in particular scatter bad years across a run rather than clustering them, which is exactly the pattern that ruins real retirements.

Fixed return. One growth rate, compounded. Still the default in most bank and broker calculators, and worth naming because it cannot express sequence-of-returns risk at all: the same average with the losses at the front is the difference between a plan that works and a plan that does not, and a fixed-return projection prices both identically. If a tool asks for an expected return and returns a single line, that is what it is doing.

And "success rate" is a sample statistic, not a forecast. It says what fraction of the scenarios the model generated ended above zero. It carries no information about how badly the failures failed — a plan that ran out in year 29 and one that ran out in year 12 are the same failure to the counter — and it assumes you would have kept spending on schedule while watching the balance fall, which nobody does. Some of these tools let you model the reaction (a spending floor, a guardrail, a variable-percentage withdrawal); that setting changes the headline number more than the allocation slider does.

Which history, and whose

Ask two questions of the dataset before you read any output from it: where does it start, and which country is it about.

The historical simulators here run on Robert Shiller's series, beginning January 1871 — S&P 500 prices and dividends from 1926 with Cowles index data before that, 10-year US Treasury yields for bonds, and the US CPI for inflation. It is the deepest free dataset in existence for this purpose and it is entirely American, which is the quiet limitation of the whole category. A reader in Berlin or Toronto gets a number computed from a century of American exceptionalism and American inflation. Portfolio Charts is the deliberate exception: it starts only in 1970, but it computes in any of twelve home currencies and deflates by that country's own CPI, which is a different and in some ways more useful trade. Backtest by Curvo is the European exception on the fund side. It covers 1,306 European-domiciled UCITS funds mapped onto 504 index series, the oldest of them starting at the end of 1969, all held in euros and displayed in euros, pounds or dollars — no Swiss franc or Swedish krona base — with Belgian CPI as the only inflation series. It is monthly, funds only, and free with no paid tier.

Backtesters that price real funds have a second, shorter horizon, and they deal with it in three different ways. Portfolio Visualizer and PortfoliosLab stop where the funds stop. Portfolio Visualizer's asset-class series reach back to 1972, but an individual fund only to its own inception — Vanguard's 500 Index fund to September 1976 — and its free tier truncates ticker backtests to the last ten years and 15 assets, so a free account cannot reach the start of the provider's own data. PortfoliosLab runs on daily prices and extends nothing: a portfolio begins on the day its youngest holding launched, and the free plan sees five years. Curvo and testfolio fill the gap instead, and the filling is a model, not a record. For Curvo's index funds and ETFs, the years before launch are the index with a twelfth of the fund's ongoing charge taken off each month, assuming perfect tracking. testfolio goes further back — daily data to 1885 for the US market — by splicing a modelled segment onto 105 of its preset fund series, and for its leveraged proxies the assumed financing and swap costs decide the answer; nothing in the result marks where the join is. Its free tier keeps the full history and withholds saving and downloads; Pro is $15 a month.

Two smaller things that silently move results: whether the tool works in real or nominal terms (all the credible ones default to real, but confirm it before comparing outputs across tools), and whether returns are net of fund fees and taxes. Portfolio Charts states plainly that it ignores both. That is the right choice for a portfolio-behaviour question and the wrong assumption to carry into a spending plan.

What it models besides returns

This is where the free tools stop and the subscriptions earn their money. None of it is about markets.

Account type. A dollar in a traditional 401(k), a Roth IRA and a taxable brokerage account are three different dollars, and a simulator that holds one undifferentiated portfolio is answering a simpler question than the one you asked. Withdrawal ordering, Roth conversion ladders, required minimum distributions and the interaction between conversion income and everything below are the core of what Boldin's PlannerPlus tier and ProjectionLab's paid tier actually sell — and what FIREproof adds to cFIREsim's historical method, with ACA premium credits projected by ZIP and the subsidy cliff flagged.

Social Security and pensions. Claiming age, spousal and survivor benefits, and the portion of the benefit that becomes taxable. A stream that starts at 67 changes the shape of the first decade of withdrawals, which is precisely the decade sequence risk operates in.

Spending that is not a straight line. Real spending is lumpy — a roof, a car, tuition, one-off travel in the first retired decade, care costs in the last. A tool that accepts a single annual figure inflated at CPI is smoothing away the shocks that the simulation exists to test.

Health insurance before Medicare. For anyone retiring before 65 in the US this is frequently the largest controllable line in the plan, and it is coupled to the tax model rather than separate from it: income recognised from a conversion or a taxable sale raises modified AGI, which sets the marketplace premium subsidy, and later sets the Medicare IRMAA surcharge on a two-year lookback. Optimising taxes without those couplings produces a confidently wrong answer.

State taxes. Federal-only modelling is common and is a meaningful error for a retiree in California or New York, and a different one for a retiree in Florida.

Where your financial picture ends up

This category asks for more than any other in the catalogue — balances by account, salary, pension, spouse, mortgage, expected inheritance. Where that lands is a real differentiator here, not a compliance footnote.

Three shapes exist. The unauthenticated calculators hold nothing: cFIREsim, FI Calc and Portfolio Charts take inputs in the browser and forget them, at the cost of rebuilding the scenario every visit or keeping the URL. Backtest by Curvo needs no account either, and keeps saved portfolios in your browser unless you sign in with Google, which moves them to its server. ProjectionLab keeps plans in its own cloud database, exports the whole account to a JSON file that restores into the app, and never links a real account; the self-hosted build it once sold to Lifetime customers has been sunset. Boldin sits at the other end, with optional account aggregation through a bank-linking provider, which is genuinely convenient and means live credentials to your institutions are in the chain.

None of these is the wrong answer. The point is that it is a choice you should make on purpose, because the aggregating tools are the ones that make it easiest not to.

Two jobs, bought by different people

Retirement is a cash-flow question: will the money last, given these accounts, this tax treatment, this spending and this claiming age. The output is a survival rate or a set of ending balances over a horizon, and the work is in the plumbing — taxes, benefits, lumpy expenses, the order in which accounts are drained. cFIREsim, FI Calc, FIREproof, Boldin and ProjectionLab answer this.

Allocation is a portfolio-behaviour question: what does this mix of assets do over long horizons — its drawdowns, its dispersion of outcomes, how long it took to recover, how a different weighting would have behaved. It has no opinion about your accounts, and it does not need one. Portfolio Visualizer, PortfoliosLab, testfolio, Backtest by Curvo and Portfolio Charts answer this.

They are bought by different people for different decisions, and a ranking that puts a withdrawal simulator above a backtester is comparing the fit of a coat to the fit of a pair of shoes. The overlap is one number — a static withdrawal rate a portfolio sustained historically — and both sides compute it, from different data and for different purposes. Decide which question you have before you read a list.

What any of this costs

Say it plainly, because the category is heavily monetised and the free end is genuinely good: cFIREsim, FI Calc and Portfolio Charts cost nothing, and for the question "what withdrawal rate did this allocation survive historically" they are not a compromised version of anything. Start there. Backtest by Curvo is free too, with no paid tier at all, for the allocation question in European funds.

The paid tier is not selling better simulation. It is selling everything listed two sections up, plus saved plans and scenario comparison. The cheapest step is FIREproof at $10 a month or $89 a year: the same historical-cycle engine as cFIREsim, with withdrawal order, Roth conversions, 72(t), bracket-filling withdrawals and ACA subsidy optimisation added — and all of them on Pro, since its free tier holds one saved simulation and none of the tax modelling. Boldin's PlannerPlus is $144 a year, billed annually after a 14-day trial with no month-to-month option; its free Basic tier takes 100-plus inputs but withholds Monte Carlo, tax projections and saved scenarios. Boldin also sells a one-off CFP plan review at $3,200, which is advice rather than software and outside what this listing covers. ProjectionLab is $129 a year for Premium and $549 for the Pro tier aimed at advisers, with a free tier that runs Monte Carlo and historical backtests on one saved plan but has no tax engine. On the allocation side, Portfolio Visualizer charges $30 a month for Basic and $55 for Pro, both billed annually, which raise the asset cap to 150, unlock the full history and — on Pro — permit commercial use; PortfoliosLab's Plus is $10 a month and lifts its five-year cap; testfolio's Pro is $15 and buys saving and downloads rather than history. All prices checked in September 2026. Pralana sells one tax-heavy model two ways, each at $99 a year plus a $30 sign-up fee as of October 2026: Pralana Gold as an Excel workbook, Pralana Online in a browser.

The honest framing: if your plan is one portfolio and one spending number, the free tools finish the job. If your plan has two people, four account types, a pension, a conversion strategy and eight years to bridge before Medicare, a spreadsheet is the alternative to the subscription, and the subscription is cheaper than the spreadsheet.

Software, not advice

Worth stating here because the niche is saturated with free tools whose actual product is your phone number. A calculator belongs in this listing if you supply the assumptions, it does the arithmetic, and it shows enough of its method for you to disagree with it. A free dashboard that aggregates your accounts in order to route you into a conversation with an adviser charging a percentage does not, however good the charts are — Empower Personal Dashboard was excluded on exactly that ground. The same line applies to these pages: they describe what each tool computes and on what data, and not what rate to withdraw, what to hold or when to claim.