Why do you use a 15-year benchmark instead of a longer timeframe?
A few reasons. Some practical, and some mathematical.
From a math perspective, 15 years is an appropriately long timeframe to bypass short-term trends and still leave enough data points in the available database to be statistically significant. It also keeps bubbles or crashes right in the middle of a data set from unfairly influencing too many results.
From a practical perspective, 15 years is long enough to be considered “long term” by most reasonable investors without watering down the real-life uncertainty involved. Most 25-year-olds do not expect to wait well beyond 40 to see their expected return finally materialize, so numbers quoted over extremely long periods sound reassuring but are actually not so helpful on normal investing timeframes.
Your 70% return numbers do not match my same calculations using standard deviations. What gives?
The Target Accuracy calculator does not use standard deviations. Instead, it actively filters out the bottom 15% and top 15% of known real-world returns. This is important because even though standard deviation calculations are a lot easier to do, portfolio returns in the real world do not follow a normal distribution pattern required for those SD numbers to accurately model portfolio performance. Study the Annual Returns calculator for different portfolios, and you’ll see what I mean.
Why do you set the bar at 70%? Why not higher or lower?
I’ve played with different filter levels, and 70% seems to strike the right balance between representing a large enough number to represent the most common outcomes but small enough to effectively filter out enough outliers. For reference, at 15 years it currently crops the top 5 and bottom 5 outcomes.