The core claim
Many planning tools start by picking a single "conservative" return (~7%), far below the historical average (~10%) and projecting one future. Appleseed takes what we believe is the correct approach: start from observable history to set a baseline, then use Monte Carlo to model uncertainty and sequence risk explicitly.
It's good to be conservative, but that needs to be balanced with the opportunity cost of being potentially overly conservative. We have decided to be what we believe is realistic with our estimates, and use Monte Carlo to understand the risk of large downsides.
Note: Appleseed's default assumes a 100% U.S. stock portfolio, which reflects how many younger investors are actually allocated. If your portfolio includes bonds, international, or cash, select a different preset (like JPM Balanced 60/40) or set custom returns - your expected return and risk level will adjust automatically.
Baseline Comparison (1976-2025)
Historical Returns: Arithmetic Mean
Below is the arithmetic mean across trailing periods and specific decades. The green line represents the ~50-year average (10.28%), showing that our 10% default is well-supported by multi-decade data.
Trailing Arithmetic Mean (Ending 2025)
Decade-by-Decade Mean
Using the 10% Default
We use 10% as the default nominal return for U.S. Equities because it is a round number that sits slightly below the historical 50-year arithmetic mean (10.28%). It serves as a neutral starting point rooted in data rather than pessimism.
User Control: This is a default, not a mandate. You can adjust this assumption in your plan settings at any time at Assumptions > Market Assumptions to see how a more conservative outlook affects your probability of success.
Why 50 years?
We look at the last 50 years of returns (1976-2025) because this period encapsulates the modern market structure (index funds, global capital mobility, etc).
We think the data suggests a structural shift in valuations, though this is actively debated - some analysts expect mean reversion, which would imply lower future returns. Our view is that even if valuations partially revert, anchoring to 50 years of data plus Monte Carlo is more honest than baking that belief into a single pessimistic point estimate.
Why 6-7% "seems reasonable" in many tools
In deterministic spreadsheets and FIRE calculators, a low return assumption is often used as a crude substitute for risk modeling. If the model can't represent dispersion, drawdowns, or bad early sequences, the only knob people have is to push the single-point return down until the projection "feels safe."
What institutional forecasts are actually saying
Many major financial institutions (like J.P. Morgan, Schwab, and Vanguard) publish “capital market assumptions” that predict returns for the next 10 years. These conservative numbers are often used as the "responsible" default in planning software.
However, these forecasts have a history of being overly conservative. For decades, they have generally projected returns well below what the market actually delivered.
22-Year Reality Check (Average 2004-2025)
Data source: Comparison of J.P. Morgan Long-Term Capital Market Assumptions (US Large Cap projections) vs. actual S&P 500 returns from 2004-2025 (22 years was the furthest back data available). Note: this 22-year window (2004-2025) includes an extended bull market, but also the 2008 financial crisis, the 2020 COVID crash, and the 2022 bear market.
Bottom line
This is a genuine debate. Our view is that the most honest planning approach is:
- Anchor the baseline to observable long-run outcomes (and show the window you chose).
- Model uncertainty with Monte Carlo (volatility + sequence risk).
- Make conservatism an explicit knob, not an implicit default.
Try running your own plan with Appleseed's Advisor-Grade Monte Carlo in 5 minutes →