Historical Backtesting: Test Your Retirement Plan Against Nearly 100 Years of Real Markets

Monte Carlo simulation asks “what might happen?” by generating thousands of random scenarios. Historical backtesting asks “what would have happened?” It runs your exact retirement plan through every real market period since 1928 (the Great Depression, stagflation, the dot-com crash, 2008, COVID) and shows how your plan would have fared in each.

The returns are actual, not assumed: the S&P 500 and 10-year Treasury total returns (price change plus dividends or interest) for each year since 1928. They are nominal, before inflation. cinder.fi inflates your spending at your plan’s inflation rate, the same as your main projection.

How It Works

cinder.fi’s backtester takes your full retirement plan (savings, spending, tax optimization, government benefits, account types) and replays it through overlapping historical windows.

For a 30-year retirement, the first window uses 1928-1957 market returns, the second uses 1929-1958, and so on, up through the most recent complete window. Each window runs your complete projection engine: tax calculations, CPP/Social Security benefits, RRSP/401(k) withdrawals, RRIF minimums, and OAS clawback. It is the same engine as your main projection.

You get nearly 70 tests of your plan, each against a real sequence of market returns.

What Gets Tested

Your backtest uses your whole plan:

• Account balances and contributions: RRSP, TFSA, 401(k), IRA, Roth, non-registered, LIRA, FHSA, HSA
• Tax optimization: RRSP meltdown, Roth conversion, TFSA maximization, SBLOC/PLOC
• Government benefits: CPP/QPP, OAS, GIS, Social Security at your chosen claiming ages
• Real estate: rental income, mortgage payments, downsizing
• Spending strategies: Fixed, Constant Dollar, Guyton-Klinger, Variable Percentage Withdrawal

The only thing the backtest overrides is investment returns. Instead of your assumed rate, each window applies the actual historical stock/bond returns for that period, blended according to your chosen allocation.

Three Backtesting Modes

Retirement Only (Decumulation)

The default mode. Your accumulation phase uses your assumed return rate, and historical returns are applied only during retirement. It answers: “Given my projected retirement portfolio, how would different market environments affect my retirement?”

This mode produces the most windows (nearly 70) since each window only needs 30 years of data.

Full Lifecycle

Historical returns apply to both your saving years and retirement. If you’re 15 years from retirement with a 30-year retirement duration, each window needs 45 years of contiguous data. This produces fewer windows but tests your complete plan end to end, including whether a crash during your saving years would leave you with too little at retirement.

Accumulation Only

Tests only your saving years with historical returns. Useful for younger savers who want to see how different markets during their working years would affect their portfolio at retirement.

Reading the Results

Success Rate

The headline number is your historical success rate: the percentage of all tested windows where your portfolio survived through retirement. A rate above 90% means your plan survived the Great Depression, stagflation, the dot-com crash, and nearly every other rough period in modern market history.

Historical Periods Timeline

The timeline groups results by historical era, from the Gilded Age through the Digital Age. Each era shows:

• Survival ratio: how many windows in that era survived (e.g., “8/10”)
• Average annual return: the geometric mean (nominal) return for windows starting in that era
• Individual windows: click any window to see a portfolio trajectory chart with historical event markers

Depleted windows are collapsed into a summary line (e.g., “6 of 10 depleted, yr 22-28”) to keep the focus on survived windows. Expand the summary to see individual depleted results.

Spending Strategy Comparison

The Spending Strategies tab compares how four withdrawal approaches would have performed across all historical windows:

• Fixed (Inflation-Adjusted): the classic 4% rule approach, with constant real spending regardless of portfolio performance
• Constant Dollar: fixed nominal spending with no inflation adjustment, so purchasing power declines over time
• Guyton-Klinger: spending adjusts with portfolio guardrails, cutting in bad years and raising in good ones
• Variable Percentage Withdrawal (VPW): spending scales with portfolio size and remaining years

If your Fixed strategy success rate is below 90%, switching to Guyton-Klinger or VPW often pushes it above 95%, because adaptive spending protects against bad sequences.

Allocation and Duration

Use the controls to test different scenarios:

• Stock/Bond Allocation: slide between 0% and 100% stocks. Historical data suggests 60-80% stocks has the best risk-adjusted outcomes over 30-year periods, but your optimal allocation depends on your spending rate and flexibility.
• Retirement Duration: test 20, 25, 30, 35, or 40-year windows, or use your plan’s actual projected duration. Longer durations naturally produce lower success rates.

Changes take effect immediately; the backtest re-runs as you adjust.

Historical Backtesting vs. Monte Carlo

Both tools test your plan’s resilience in different ways:

Use both. Historical backtesting checks your plan against known worst cases. Monte Carlo simulation tests it against scenarios that haven’t happened yet.

Practical Tips

1. Don’t chase 100%. A 100% historical success rate may mean you’re spending too conservatively. Many retirees with 100% success rates die with 3-5x their initial portfolio. A 90-95% rate with higher spending may fit your goals better.

2. Check the worst period. If your plan fails, look at when it fails. A plan that fails only in 1929-era windows is in a different position from one that fails in 1970s stagflation.

3. Test spending flexibility. If your Fixed strategy rate is concerning, check Guyton-Klinger or VPW. Willingness to cut spending 10-15% during crashes often raises success rates by 10-20 percentage points.

4. Returns are nominal. cinder.fi’s backtest uses each year’s actual returns before inflation and inflates your spending at your plan’s inflation rate. A “5% average return” in the backtest is a nominal 5%, the same kind of number as the return assumptions in your plan.

5. Use both modes. Run Retirement Only for the most windows and the clearest retirement-phase picture. Run Full Lifecycle to see how pre-retirement market risk carries into retirement outcomes.