How risk in dollars is computed

The allocation explorer answers one question: for a given mix of stocks and bonds, what did good and bad years look like, in dollars for your portfolio? It appears in the investment plan builder, on your plan page, on each allocation model, and as a free public tool. It shows history. It does not say which mix to hold; you choose your own range.

The data

The figures use the same public annual series as Ironlake's historical replay and Monte Carlo stress view: Robert Shiller's U.S. data for stocks (S&P 500, dividends reinvested), 10-year U.S. Treasury bonds, and inflation (CPI), from 1928 to 2022. The U.S. series run January to January, so a "one-year period" in the explorer starts in January and ends the following January. It is close to a calendar year but not identical: for example, the 2022 figure for U.S. stocks is about -12%, while the calendar-year 2022 decline was about -18%.

The mix

A mix such as "60% stocks, 40% bonds" means 60% in U.S. stocks and 40% in 10-year Treasuries, rebalanced back to those weights at the start of every one-year period. The explorer does not include cash, international stocks, or other asset classes, so a model that holds them is approximated by its stock share. Figures are before taxes, fund expenses, and advisory fees.

What each figure means

• Worst, typical, and best one-year period: the lowest, median, and highest one-year return of the mix across the 95 periods, multiplied by the portfolio value you enter.
• Lost money in N periods: how many of the one-year periods had a negative return.
• Grew X% a year after inflation: the compound annual return over the whole span after dividing out inflation.
• Worst fall from a high: the largest decline from one year-end value to a later lower year-end value, with the year the prior high was regained. The explorer shows this both before and after inflation, because inflation can turn a modest nominal decline into a long real one (for example, an all-bond mix lost more than half its purchasing power from 1940 to 1981).
• Named stretches: the cumulative return of the mix over well-known bad periods (1929 to 1932, 1973 to 1974, 2000 to 2002, 2008, 2022).

Falls within a year were deeper

Year-end values smooth over what happened in between. The explorer therefore shows a separate reference table of well-known peak-to-trough declines for U.S. stocks alone, measured on daily closing levels of the S&P 500 price index (dividends excluded). For example, U.S. stocks fell about 57% from October 2007 to March 2009 and about 34% in five weeks in early 2020, even though the 2020 one-year figure is positive. These reference figures are for display only and do not enter any calculation.

Limitations (read these)

• Past results do not predict future results. The 1928 to 2022 record is one history of one country's markets.
• Annual data understates short-term losses. A portfolio can fall much further within a year than any year-end figure shows; the reference table gives a sense of how much.
• Two assets only. U.S. stocks and 10-year Treasuries stand in for every holding, and rebalancing is assumed to happen exactly once a year.
• Before taxes and costs. Real portfolios pay fund expenses, fees, and taxes on income and rebalancing.
• Not advice. Ironlake shows the history and the arithmetic; the range you choose is your own decision.

Try it with your own portfolio value