Return and volatility assumptions: what to put in a retirement plan and why

Every chance-your-money-lasts figure rests on two kinds of number you type in: how much markets will earn on average, and how far a typical year strays from that average. Runway's Market Assumptions step has four fields for them: an expected return and a volatility for stocks, and the same two for bonds. Nobody knows the right values, so the useful questions are what history supports, what today's bond yields say, and how much each number actually matters. This guide answers all three.

What do expected return and volatility mean?

The expected return is the average yearly gain, and Runway asks for it before inflation (in nominal terms) and then removes the expected inflation rate, so everything inside the engine is in today's dollars. With the default 2.5% inflation, the defaults of 10.2% for stocks and 5.5% for bonds become 7.5% and 2.93% a year after inflation. (This post explains why the difference between nominal and real matters so much.)

The volatility is the standard deviation of a year's return, a measure of how widely years scatter around the average. Runway's stock default of 16% means that, if returns follow a bell curve, about two years in three land within 16 points of the average: between a loss of 8.5% and a gain of 23.5% after inflation. Each simulated year in the Monte Carlo stress test is drawn from a bell curve built from the average and the volatility you enter, with stocks and bonds moving slightly opposite to each other by default (a correlation of −0.10).

What does history say?

Runway's historical backtest uses annual real total returns for U.S. stocks and 10-year bonds from Robert Shiller's public dataset, 154 years in all. Here is what that same series says about the four numbers:

The two averages differ for a reason that matters here. The arithmetic average is what you enter; the geometric average is what an investor actually compounded at. The gap between them is roughly half the variance: for stocks, half of 17.94% squared is 1.61 points, and the actual gap is 1.58. A volatile investment earns less than its average suggests, and the more it swings, the bigger the gap.

The recent decades were kinder than the long run. Since 1950, stocks averaged 9.23% with 16.72% volatility, and since 1980, 9.91% with 15.40%; bonds averaged 2.08% since 1950 and 3.92% since 1980, with volatility above 9% and 10%. Averages also hide how different any one stretch can be. Across all the 30-year windows in the series (125 of them), the annualized real stock return ranged from 3.3% at worst to 10.4% at best, with a median of 6.8%. Over 10-year windows the range was −4.2% to 17.4%, and over single years it was −39.3% to 53.2%.

What do today's bond yields say?

Stocks have no yield that tells you what comes next, but bonds do. Treasury publishes the daily real yield curve, the yield on inflation-protected Treasury bonds after inflation. On October 6, 2026, it was 2.66% for 5-year, 2.91% for 10-year and 3.35% for 30-year bonds. At 2.5% inflation, 2.91% real is 5.48% nominal. So Runway's default bond return of 5.5% before inflation, or 2.93% after it, sits right on where a 10-year inflation-protected Treasury is yielding today, and just above the 2.78% history average. (Runway's default was 3.5% after inflation until October 6, 2026; it was lowered because it was above these yields.) Moving the bond return in the engine changes the sample couple's success rate modestly, because the plan leans on stocks: at 2.0% it is 98.6% at $95,000 of spending and 80.7% at $110,000; at the 2.93% default, 99.4% and 84.9%; at the old 3.5%, 99.5% and 87.2%. If your bonds are not Treasuries (corporate or municipal funds, say), their yield can differ from these.

How much does each number matter?

Here is the sample couple (Pat 65, Alex 64, $1.46 million, plan through 95) with one number changed at a time. Start with stock volatility, holding the 7.5% average fixed:

Now the expected stock return, holding the 16% volatility fixed:

Two things stand out. First, both numbers matter, and the closer a plan is to the edge, the more they matter: at $95,000 the whole range moves success by 3.5 points, while at $110,000 it moves it by 23, with each point of stock return worth 7 to 10 points. Second, volatility costs you even when the average is unchanged. With no volatility at all, the sample plan would end with $1,727,686; with the default volatility, the median ending balance is $1,483,604, which is $244,082, or 14.1%, less. Losses hit a smaller base while withdrawals keep coming out, so a bumpy ride to the same average leaves less.

So what should I enter?

There is no correct answer, so treat the three fields as a range to test. Here are three sets, with the nominal numbers to type in (at 2.5% inflation) and what each does to the sample couple:

The defaults and a set built from 154 years of history land close together, which says the defaults are not unusually hopeful. The cautious set is the one to take seriously as a test: it cuts the stock return by 2 points and the bond return by 0.9 and raises both volatilities, and the $110,000 plan falls from 85% to 58%. If a plan only works at the default numbers and collapses under the cautious ones, it has little margin; if it holds up under both, the exact inputs matter less. Two practical rules follow. Enter returns after fund costs, because Runway has no separate fee field. And change one assumption at a time when you test, so you can see which one your plan is sensitive to.

What does this leave out?

• The bell curve. The simulation draws each year independently from a bell curve. Real markets have fatter tails and sometimes run in streaks, which is why Runway also runs the historical backtest and why the order of returns matters.
• A fixed correlation. The stock and bond relationship is set at −0.10. In the Shiller series it was +0.16 over the whole period, a number you can't change in the app.
• One country, one dataset. The history is U.S. only, and past returns don't promise future ones.
• Your own mix. The blended return depends on how much is in stocks (see the asset allocation guide); a lower stock share makes the bond numbers weigh more.

How were these numbers computed?

The history figures are computed from the annual real returns for U.S. stocks and 10-year bonds in Robert Shiller's public dataset, as stored in Runway's engine (1872 through 2025); the windows are every 1-, 10- and 30-year stretch in that series. The Treasury yields are from its daily real yield curve rates for October 6, 2026. The sample-couple results are Runway's Monte Carlo engine: married filing jointly, Social Security claimed at 69 and 68, plan through 95, 60% stocks and 40% bonds, stocks at 7.5% and bonds at 2.93% after inflation unless changed, volatilities of 16% and 6%, correlation −0.10, 2,000 simulated markets with seed 2026, and long-term-care costs left out. Dollar amounts are in today's dollars, and nominal figures assume 2.5% inflation.