$MART DEBT Coach · Market Drop Wins research · companion report ·
A rule you can commit to before the next decline
At record valuations, the decade ahead offers a much worse reward for its risk, and that affects almost every stock investor's whole portfolio, not just new money. But valuations can't tell you when a decline starts. So the useful question is how to adjust your stock allocation without predicting anything. From the Dec 1999 peak, $100,000 left alone was worth about ten years later; halving stocks whenever the trend broke left about . This report tests a simple, pre-committed rule over 100 years of US data, plus Canada and Japan: hold stocks, but move to cash only when valuations are extreme and the price trend breaks, or just halve stocks, which is easier to live with.
What we found
The rule
Written down in advance
Check once a month, at month-end. Use the index's total-return level (dividends included) and its 10-month simple moving average.
Is the market expensive? Shiller CAPE at or above its 90th percentile of history to date.
If expensive and the level closes below its 10-month average: move the stock allocation to cash (T-bills, a high-interest savings account or a cashable GIC). The gentler version moves half: 100/0 becomes 50/50, and 60/40 becomes 30/70.
Otherwise stay, or go back, fully invested. That includes whenever the level closes back above its average.
Signals are known at month-end and applied to the next month. CAPE is lagged one extra month for publication delay, and every switch costs 0.1%. The “trend only” comparison drops step 2.
The practical dial
All-or-nothing, or halve it?
Big switches are hard for both investors and advisors to adopt. The same signals with a smaller response: when the rule fires, cut stocks by 50% or by 100%. Each row is compared with its own buy-and-hold. The 60/40 rows put the freed money into bonds. Rebalanced monthly; 0.1% per switch.
Crashes
What each approach lost in the big declines
A century of results
Growth of $1, US total market, 1926–2026
Does it still work?
By era, including after the rules were published
The trend rule was popularised by Faber (2007) and the valuation-plus-momentum idea by Asness, Ilmanen and Maloney (2017). Results after publication are the honest out-of-sample test.
Robustness
A plateau, not a lucky setting
Every combination of valuation threshold and moving-average length, plus a rate-adjusted valuation signal (Excess CAPE Yield). A result that holds across neighbouring settings is less likely to be a fluke of one choice.
Behaviour
Where investors are most likely to abandon a rule
Other markets
Canada and Japan
Using it
For investors and advisors
Treat it as insurance, not a return engine. Its value is a smaller worst case, which keeps an investor invested through the next decline. Expect to trail a rising market for years.
Choose the goal first. For someone who wants maximum long-run wealth and can sit through −50% or worse, buy-and-hold remains hard to beat. For someone already financially independent, or who would sell in a panic, a rule that caps the damage is often worth its cost.
Use a tax-sheltered account. In a taxable account, each exit realizes capital gains. Keep the rule's switching inside an RRSP or TFSA.
Pre-commit in writing, including the re-entry. Most of the damage from timing comes from never getting back in. The rule re-enters mechanically.
For investors who borrow to invest: the same logic argues for not adding leverage while valuations are extreme and the trend is weak, and for having a written plan to add it after a decline.
Compliance: anyone giving this to clients should get supervisor pre-approval for client-facing material first.
Method and limits
How this was built
US: CRSP value-weighted total market (month-end) and 1-month T-bills from the Ken French data library, 1926–Jun 2026. CAPE from Robert Shiller's data, ranked in real time (expanding window from 1881).
Canada: S&P/TSX total return (StatCan to 2001, XIC.TO after) vs 3-month T-bills, 1957–2026. The valuation signal is US CAPE, because there is no complete Canadian series. After-tax results use 2026 Ontario rates at a $200k income, with gains realized on each exit.
Japan: Nikkei 225 month-end (FRED, price-weighted), with approximate dividends and cash rates (Jordà–Schularick–Taylor). The valuation signal is an absolute dividend yield below 1%.
Several rules and 23 settings were tested, so every “best” result carries multiple-testing risk. Prefer the plateau over the peak.
Not modelled: fund fees, bid-ask spreads beyond 0.1% a switch, and taxes in the US results.
Main report (the valuation evidence): What Would You Have Paid in 1999 to Know?. Reproduce with uv run python scripts/run_vt.py && uv run python scripts/build_vt_report.py in ~/projects/cape-timing.