Portfolio Diversifier Calculator

How the diversifier calculator works

1. Replay history compares US stocks plus your chosen diversifier with the same stock allocation plus Treasury bills, and with stocks alone. Choose any inclusive period from 2012 through 2025. The starting example invests $100,000 with 80% stocks / 20% AQMIX over that full period, with extra costs at zero; it illustrates a historical result, not a recommendation.
2. Annual rebalancing restores the chosen weights at the start of each year. With it off, holdings drift. The replay compounds each annual asset return; it does not average the assets’ CAGRs.
3. Retirement spending is withdrawn proportionally before each year’s return, increasing at your chosen constant rate. Balances stop at zero and unmet spending is reported. Reversing the paired annual observations creates a hypothetical sequence test.
4. Extra costs default to zero. Holding costs and hypothetical tax drag have separate annual rates for stocks, the selected diversifier and cash. After each year’s market return, holding costs reduce each sleeve’s value, then tax drag reduces the remaining value. Fund expenses already embedded in source returns are never deducted again.
5. One-way trading costs apply to initial purchases, proportional withdrawal sales and both sides of annual rebalancing, paid from portfolio capital. Initial invested value equals starting cash divided by (1 + trading rate). Spending is limited to sale proceeds after trading costs. No final liquidation is assumed. Cash uses its own holding and tax inputs, and the same trading rate applies to all comparisons.
6. The headline compares the blend’s compound return and annual-return volatility with the same stock allocation plus cash after costs in a separate no-withdrawal replay. It identifies improvement, underperformance, or a tradeoff; it does not establish retirement safety. Differences below 0.005 percentage points are treated as similar.
7. Explore assumptions calculates weighted arithmetic return and portfolio variance: (1−w)²σs² + w²σd² + 2w(1−w)ρσsσd. Rates and volatility are decimals in the formula; w is the diversifier weight.
8. Compare the blend with the same stock allocation plus cash before calling a diversifier useful. Lower volatility can come with lower return, higher cost, or risks the variance formula does not capture.

Worked example

With 80% stocks at an assumed 8% arithmetic return and 18% volatility, plus 20% cash earning 3% at zero volatility, modeled return is 7% and volatility is 14.4%. A different diversifier has to justify its extra risk or cost against that cash blend.

Assumptions and boundaries

• Core series available for 2006–2025; the seven-diversifier replay uses 2012–2025. Calendar years, USD nominal. Stocks include dividends; 10-year Treasury returns include income and price changes. Cash uses the source’s average 3-month T-bill rate as a return proxy, not a traded cash index. Gold is the source’s price-return series. No fund fees or taxes.
• 2012–2025 is the longest shared set of full calendar years for these selected live-fund proxies and the core data. BTAL launched September 13, 2011; its partial launch year and incomplete 2026 are excluded. No missing years are filled or backtested.
• AQMIX uses Class I NAV total returns; DBC, XLP and BTAL use ETF market-price total returns. Distributions are reinvested; fund operating expenses and realized internal trading and financing effects are already reflected. ETF returns can differ from issuer NAV returns because of premiums, discounts and pricing conventions. The source series do not deduct investor taxes, commissions or bid/ask costs; optional extra-cost inputs adjust the replay separately. Checked October 21, 2026.
• CAGR, annual-return volatility, worst year and year-end max drawdown use a separate portfolio with no withdrawals, including its own holding, tax-drag and trading costs. Actual dollar costs reported beside spending come from the withdrawal portfolio. Volatility uses sample standard deviation (n−1); correlation uses matched annual observations. A single year has no estimated volatility or correlation.
• The model presets are teaching inputs, not forecasts or historical calibrations. The diversifier fee reduces only that allocation’s assumed arithmetic return. Cash is modeled with a constant return and zero volatility.
• The approximation g ≈ μ − σ²/2 illustrates volatility drag. It is not an exact expected compound return, especially for extreme assumptions. The historical replay uses actual compounding instead.

Important limitation: Annual data hide intrayear crashes, including March 2020. Optional costs and tax drag are sensitivity assumptions, not actual tax calculations or execution estimates. The model excludes tax lots, tax-loss credits, account-specific rules, withdrawal taxes, final-sale costs, contributions and actual inflation; withdrawal growth is an assumption. Treasuries are a 10-year proxy, not a broad bond fund. Live fund proxies are AQMIX, DBC, XLP and BTAL. Their strategies and expenses can change; the common window excludes 2008. A 14-year sample cannot establish future expected returns or a safe withdrawal rate.

Sources and review

Methodology and links reviewed October 21, 2026. This educational estimate is not financial, tax, legal, investment, or lending advice.

• AQMIX · AQR 2022 prospectus (2012–2015)
• AQMIX · AQR 2026 prospectus (2016–2025)
• DBC · ETFreplay annual total returns
• XLP · ETFreplay annual total returns
• BTAL · ETFreplay annual total returns
• Aswath Damodaran, NYU Stern · Historical returns
• AQR · A Century of Evidence on Trend-Following Investing
• Vanguard · Rebalancing and risk control
• Read FireFolio’s companion article