"Sell when pricey" backtest

In One Sentence

I tested the rule "sell when the price runs too far above its 10-year average" on decades of data from the US, Japan and Taiwan: whether I sold everything, loosened the threshold or only trimmed, it lost to holding in all three markets.

Why It Matters

I believed in this rule for many years, and looking back, it was spot-on in 2000 and 2008. But in July 2024 it told me to sell everything, and by October 2026 Taiwan's market had doubled from where it stood when the signal came. A market can stay in "too expensive" territory for years, and people who sell often end up waiting a decade or more. When a method turns out to be wrong, change it — that's the attitude I want to leave my family.

One Number

Selling everything when the price topped 150% of its 10-year average, and buying back only once it fell back to the average, left you with less wealth than holding: 97% less in the US, 78% less in Japan and 64% less in Taiwan. Only a complicated version — "sell only when the price breaks below its moving average" — came out slightly ahead, by 4% to 11%, and it's hard to stick to for years on end (author's backtests · Chapter V).

In the Book

Section 6 of Chapter V tells the whole story, with the backtest tables. It's also why the thermometer in Chapter VI changed its purpose.

Go Further

• The timing trap "Sell when pricey" is the most common kind of market timing.
• Buying in crashes I swapped selling for adding more from new income when markets crash.
• Robert Shiller The US part of the backtest uses Shiller's long-run data.
• Long-term holding Not selling at the highs keeps compounding rolling all the way.
• Market thermometer From then on, the thermometer only tracks emotions; it's no longer a sell signal.