The Almanac · Vol IMethodologyOpen-source
Open-source methodology

Our Methodology

Most retirement calculators use a single average return and call it a day. We test your plan against 150+ years of real market history -- every crash, every boom, every sideways decade. Here's exactly how, and why it matters.

§ 01 · How it works

How we calculate your projections.

From your inputs to actionable insights in four steps.

01 · Inputs

Your Inputs

Current age, retirement age, savings, spending, asset allocation, Social Security, and other income sources.

  • ·Current portfolio value
  • ·Monthly contributions until retirement
  • ·Expected retirement spending
  • ·Social Security claiming age
  • ·Pension or other income
02 · Data

Historical Data

Two data sources: Shiller's dataset (1871–present) for Historical Bootstrap mode, or per-asset Yahoo Finance data for Parametric mode.

  • ·Shiller: annual stock and bond returns (bundled snapshot)
  • ·Yahoo Finance: per-asset CAGR and volatility
  • ·Bootstrap preserves real stock/bond correlations within each year
  • ·Choose your data source per scenario
03 · Simulate

Monte Carlo Simulation

Two simulation methods: Historical Bootstrap samples full-year tuples; Parametric generates correlated returns from per-asset distributions.

  • ·Historical Bootstrap samples real year tuples with replacement
  • ·Parametric: correlated normals via Cholesky decomposition
  • ·Bootstrap preserves fat tails and regime changes
  • ·Free: up to 200 runs; Retirement Confidence: up to 10,000 runs
04 · Analyze

Results Analysis

We analyze all scenarios to calculate success probability, safe withdrawal rates, and risk metrics.

  • ·Success rate (% of paths where money lasts)
  • ·Percentile bands (10th, 25th, 50th, 75th, 90th)
  • ·Worst-case analysis
  • ·Optimal withdrawal recommendations

Key insight: Unlike simple calculators that assume a fixed 7% annual return, Monte Carlo simulation shows you the range of possible outcomes. Markets don't return a steady percentage — they're volatile. Our approach captures that reality.

§ 02 · Core method

Monte Carlo Simulation.

The gold standard for retirement planning used by financial institutions worldwide.

Monte Carlo simulation runs your retirement plan through thousands of scenarios based on historical market data. Instead of assuming markets return a steady 7% every year (they don't), we test what happens when you retire into a bull market, a bear market, or anything in between.

How Bootstrap Sampling Works

We use a technique called bootstrap sampling. For each simulation:

  1. We randomly select a year from history (e.g., 1973)
  2. We use that year's actual stock returns and bond returns
  3. We repeat for each year of your retirement (e.g., 30 times)
  4. This creates one possible "retirement path"
  5. We run hundreds to thousands of paths to build a distribution of outcomes

Because we sample entire years together, we preserve the real correlation between stocks and bonds. When stocks crashed in 2008, bonds rallied. That relationship is captured in our simulations. Inflation is handled separately using your configured rate.

Two Simulation Modes

Historical Bootstrap

Draws full calendar-year tuples (stock return, bond return) from Shiller's dataset spanning 1871–present. Preserves real correlations, fat tails, and regime changes. Best for traditional stock/bond portfolios.

Parametric

Uses per-asset CAGR and volatility from Yahoo Finance, then generates correlated returns via Cholesky decomposition. Better for portfolios with non-traditional assets like crypto or individual stocks.

Why Not Just Use Average Returns?

Deterministic Projection

Assumes a fixed 7% annual return every year.

Pros
  • Simple to understand
  • Single clear answer
Cons
  • ×Ignores volatility
  • ×Misses sequence risk
  • ×Overconfident
Recommended

Monte Carlo (Stochastic)

Simulates many possible futures using historical return patterns.

Pros
  • Captures real market behavior
  • Shows probability of success
  • Reveals tail risks
Cons
  • ×More complex output
  • ×Requires interpretation

Understanding Sequence of Returns Risk

Two retirees with identical average returns can have vastly different outcomes depending on when the bad years occur. A market crash early in retirement (when you're withdrawing from a depleting portfolio) is far more damaging than the same crash late in retirement. Monte Carlo simulation captures this "sequence risk" that simple calculators miss entirely.

§ 02b · Try it

See it in action.

Run a simplified Monte Carlo simulation. $1,000,000 starting balance, $40,000 annual withdrawal, 30 years.

20
$2.0M$0030 years

Press Run to simulate.

Demo uses a simplified annual-returns sample. The production simulator uses Shiller's full 1871–present dataset (Historical Bootstrap) or per-asset Yahoo Finance distributions (Parametric).

§ 03 · Transparency

Our Assumptions.

Every model makes assumptions. Here are ours — and how you can adjust them.

Inflation Rate

Default: 3% annually

We default to the long-term historical average. The Federal Reserve targets 2%, but realized inflation since 1926 has averaged about 3%.

Key Considerations
  • Higher inflation erodes purchasing power faster
  • TIPS and I-bonds can hedge inflation risk
  • Healthcare inflation typically exceeds general inflation
→ Customizable in the simulator settings.

Investment Returns

Historical bootstrapping (default 7% CAGR per asset)

By default, we sample actual historical years via bootstrap. You can also switch to parametric mode and override CAGR and volatility per asset.

Key Considerations
  • Stock returns: historical average ~10% nominal, ~7% real
  • Bond returns: historical average ~5% nominal, ~2.5% real
  • Future returns may differ from historical patterns
→ Customizable in the simulator settings.

Social Security

Full scheduled benefits

We assume you receive your estimated benefit at your chosen claiming age. Benefits are inflation-adjusted. Calculations use current SSA rules for earnings tests, claiming adjustments, and COLA.

Key Considerations
  • Claiming at 62 reduces benefits by ~30%
  • Delaying to 70 increases benefits by ~24%
  • You can model reduced benefits (e.g., 77%) for pessimistic scenarios
→ Customizable in the simulator settings.

Life Expectancy

User-configured (no fixed default)

You set your planning horizon explicitly when creating a scenario. Standalone tools typically default to age 85–90. We recommend planning conservatively.

Key Considerations
  • 50% of 65-year-olds live past 85
  • Women live ~3 years longer than men on average
  • Family history and health matter significantly
→ Customizable in the simulator settings.

Spending Patterns

Constant real spending

By default, we assume your spending needs remain constant in real (inflation-adjusted) terms throughout retirement.

Key Considerations
  • Research suggests spending declines ~1% per year in retirement
  • Healthcare costs often increase in later years
  • You can model "spending smiles" or phases
→ Customizable in the simulator settings.

Tax Treatment

Flat effective rate (default 22%)

We apply a single effective tax rate to all withdrawals in projections. A separate Roth vs Traditional comparison tool models marginal brackets. No progressive bracket modeling in the core simulator.

Key Considerations
  • Actual taxes depend on income level and state
  • The default 22% approximates a moderate-income retiree
  • Use the Roth vs Traditional tool for bracket-level analysis
→ Customizable in the simulator settings.
§ 04 · Strategies

Withdrawal Strategies.

How you withdraw money in retirement matters as much as how much you've saved.

4% Rule (Constant Dollar)

Withdraw 4% of initial portfolio in year 1, then adjust for inflation annually.

Year 1: Portfolio × 0.04 then adjust for CPI
Pros
  • Simple and predictable
  • Well-researched (Trinity Study)
  • Stable income
Cons
  • Ignores portfolio performance
  • May leave money on the table
  • 30-year horizon only

Best for: Those who prioritize spending stability over optimization

Variable Percentage Withdrawal (VPW)

Withdraw a percentage of current portfolio using the IRS Single Life Expectancy Table for age-based divisors.

Withdrawal = Portfolio / Life_Expectancy[age]
Pros
  • Adapts to market performance
  • Lower failure risk
  • Accounts for mortality
Cons
  • Income varies year to year
  • Requires flexibility
  • More complex

Best for: Those comfortable with income variability who want to optimize spending

Guardrails (Guyton-Klinger)

Start with a base withdrawal rate, but adjust up/down based on portfolio performance thresholds.

Base rate ± adjustments when withdrawal rate hits guard rails
Pros
  • Balances stability and adaptability
  • Clear decision rules
  • Limits extreme outcomes
Cons
  • More complex to implement
  • Requires monitoring
  • Still has failure scenarios

Best for: Those who want some spending stability with market-responsive adjustments

Fixed Percentage

Withdraw a fixed percentage (e.g., 4%) of current portfolio value each year.

Withdrawal = Current Portfolio × Fixed %
Pros
  • Never runs out of money
  • Simple rule
  • Fully market-responsive
Cons
  • Highly variable income
  • Can drop significantly in downturns
  • No inflation protection

Best for: Those with significant flexibility or other income sources

Floor-Ceiling

Withdraw a percentage of your portfolio each year, but clamp the result between inflation-adjusted dollar floor and ceiling amounts.

Withdrawal = clamp(Portfolio × Rate, Floor × (1+i)^t, Ceiling × (1+i)^t)
Pros
  • Never runs out of money (like %)
  • Limits downside spending cuts
  • Caps excessive withdrawals
Cons
  • Income still varies within range
  • Requires setting initial floor/ceiling
  • More complex than fixed percentage

Best for: Those who want percentage-based flexibility with hard spending boundaries

Strategy Comparison

Metric4% RuleVPWGuardrailsFixed %Floor-Ceiling
Failure RiskMediumLowLowNone*None*
Income StabilityHighLowMediumLowMedium
Upside CaptureLowHighMediumHighMedium
ComplexityLowMediumMediumLowMedium
Best Use CaseStabilityOptimizationBalancedFlexibilityBounded

*Fixed percentage never "fails" because you always withdraw a % of what remains, but income can drop to very low levels.

§ 05 · Sources

Data Sources.

Our calculations are only as good as the data behind them. We use authoritative, publicly available sources.

Robert Shiller's Dataset

Updated: Bundled (annual snapshot)
visit

Annual US stock and bond returns from 1871-present (see live endpoint for exact range). The gold standard for long-term market analysis, bundled with the application.

Data Points Used
  • Annual stock total returns
  • Annual bond total returns
  • Annual CPI inflation

Yahoo Finance

Updated: Real-time (15-min cache)
visit

Historical price data for individual securities, ETFs, mutual funds, and crypto. Used to compute per-asset CAGR and volatility for parametric simulations.

Data Points Used
  • Current prices
  • Historical prices (5-year default)
  • Computed CAGR
  • Annualized volatility

Social Security Administration

Updated: Annual (current-year rules)
visit

Official benefit formulas for Social Security projections. Claiming-age adjustments, spousal and survivor benefits, and earnings test rules.

Data Points Used
  • PIA adjustment formulas
  • Early/delayed claiming rules
  • COLA adjustments
  • Spousal & survivor benefits

IRS Publication 590-B

Updated: IRS Publication 590-B Table I (current revision)
visit

Single Life Expectancy Table used for Variable Percentage Withdrawal (VPW) strategy calculations.

Data Points Used
  • Life expectancy divisors (ages 50-115)
  • Used for VPW withdrawal strategy

Data freshness: Shiller's historical return data is bundled with the application as an annual snapshot. Asset prices from Yahoo Finance are fetched in real-time with a 15-minute cache. SSA, IRS, healthcare, and federal-bracket rules reflect the current-year published tables; refresh the page to load the live freshness summary from our API.

§ 06 · Important

Known Limitations.

Every model has limitations. Understanding them helps you use the tool appropriately.

Historical Data ≠ Future Returns

Past performance doesn't guarantee future results. The future may differ significantly from any historical period. Our simulations show what would have happened historically, not what will happen. That's why we test against 150+ years of data — including the worst periods in market history.

US-Centric Data

Our primary dataset covers US markets. International diversification benefits and risks aren't fully captured. If you have significant non-US holdings, results may vary. You can override return assumptions for any asset class in the simulator.

Simplified Tax Model

Projections apply a single flat effective tax rate (default 22%) to all withdrawals. We don't model progressive brackets, state taxes, capital gains nuances, or account-type-specific treatment in the core simulator. For bracket-level analysis, use the Roth vs Traditional tool.

No Behavioral Factors

Models assume you'll follow the plan. In reality, investors often panic-sell in downturns or overspend in bull markets. Behavioral discipline matters. The Guardrails strategy builds automatic adjustments into your spending plan.

Healthcare Cost Uncertainty

Healthcare costs are highly variable and can be catastrophic. While we model average healthcare inflation, individual outcomes vary dramatically. Use the Medicare and Healthcare Bridge tools for detailed cost projections.

Black Swan Events

Monte Carlo simulations are based on historical volatility. True black swan events (worse than any historical period) aren't captured. The worst historical periods (1929, 2008) are included in every simulation.

Important Disclaimer

This tool is for educational and planning purposes only. This is not financial advice.

  • We are not registered investment advisors, financial planners, or tax professionals.
  • Results are projections based on historical data and your inputs, not predictions.
  • Consult qualified professionals for personalized advice about your specific situation.
  • You are solely responsible for your financial decisions.
§ 07 · Research

The research behind your numbers.

Our methodology is built on peer-reviewed research spanning 30+ years. The 4% rule was first established by William Bengen in 1994 and stress-tested by the Trinity Study across 75 years of market data.

Foundational Research

Determining Withdrawal Rates Using Historical Data (Bengen, 1994)

The original "4% rule" paper that established safe withdrawal rate research.

Trinity Study: Retirement Savings (Cooley, Hubbard, Walz, 1998)

Expanded Bengen's work with various asset allocations and withdrawal rates.

Irrational Exuberance (Shiller, 2000)

Robert Shiller's seminal work on market valuations and the CAPE ratio.

Monte Carlo & Simulation

Monte Carlo Simulation in Financial Planning

Overview of how Monte Carlo methods apply to retirement planning.

Bootstrap Sampling in Finance

Technical background on bootstrap methods for financial simulation.

Withdrawal Strategies

Variable Percentage Withdrawal (Bogleheads)

Community-developed approach to dynamic withdrawal rates.

Guyton-Klinger Decision Rules

The guardrails approach to managing withdrawal rates.

Spending Flexibility and Safe Withdrawal Rates

How spending flexibility impacts sustainable withdrawal rates.

Ready to see your own numbers?

Open the simulator and stress-test your plan against 150+ years of market history. Free tier available forever — upgrade to Retirement Confidence or Investing Engine when you need more power.