The complete FIRE platform is taking shape. Two practice tools are ready today.
Strategy test 001 / Retirement withdrawals / Market timing
QQQ can grow quickly and fall hard. I tested whether a signal from the broader U.S. stock market could tell a retiree when to hold QQQ and when to move to Treasury cash.
The actual 25-year return
This page matches the original study's fixed recent-history test: July 2001 through June 2026. I also have a 1926–2026 series, but QQQ did not exist before 1999 and SPY did not exist before 1993. The earlier years use research proxies, not the actual funds, and those proxy-derived results are not yet cleared for publication here.
The guard spent its first twelve months in cash because it did not yet have a full year of signals. That happened to avoid part of the technology crash. The retirement test below uses many starting months, and it did not establish a winner.
A path counted as successful only when the portfolio paid every inflation-adjusted withdrawal for 100 years and its purchasing power kept growing. Nearly every path passed under both rules. Because the small difference changed direction when the same history was arranged another way, this recent period alone does not show that either rule is better.
01 / The rule
At each month-end, the rule looks back at the previous twelve completed months. If SPY beat three-month Treasury cash in at least five of those months, the portfolio holds QQQ. If SPY won four months or fewer, it moves fully to cash. The portfolio stays in cash during its first twelve months because the signal does not yet have a complete history.
The comparison simply holds QQQ throughout. Both plans begin with $750,000, withdraw $30,000 in the first year, raise later withdrawals with inflation, use the same account-tax model, and pay 0.10% each time money is moved into or out of QQQ.
The rule does not predict a crash. It responds to whether SPY has beaten cash often enough during the year already completed.
02 / Why the two tests disagree
The first test built 5,000 century-long lives from connected eight-, ten-, and twelve-year stretches of the 2001–2026 record. The guard succeeded in all 5,000 modeled lives. Always holding QQQ succeeded in 4,957 of the same 5,000 lives—a difference of 43 modeled lives, or 0.86% of the 5,000 paths.
The second test began once in each of the 300 months, kept the 25-year record in its real order, and repeated it until each path reached 100 years. Always holding QQQ succeeded in all 300 paths. The guard succeeded in 298 of the 300 paths.
One construction favored the guard by 43 paths. The other favored always holding QQQ by two. That is an unresolved result, not a recommendation.
03 / The independent data check
I rebuilt the SPY and QQQ monthly returns from a separately licensed adjusted-price source. Treasury cash and inflation came from U.S. government data. Some individual monthly ETF returns differed from the earlier source, but the guarded and always-QQQ success counts remained exactly the same in both path tests.
No raw licensed price rows are published here. The page contains only the rules and aggregate results that cannot be used to reconstruct the source data.
This recent-history conclusion no longer depends on the previous data source.
04 / What this means for the decision
This is the easiest independently reproducible part of a much larger strategy comparison. It deliberately excludes the older proxy histories that are not yet cleared for publication on FIRE, so it cannot reproduce the full table or its long-history comparison.
The useful result is narrower: a simple SPY signal can be tested without looking ahead, and its recent-history result survives a change of data source. But the period was so favorable to QQQ that both choices nearly always passed.
The next useful test is the same head-to-head comparison across longer publishable market histories—not a live trading signal.
The full receipt
| Path construction | Guarded QQQ | Always QQQ | Difference |
|---|---|---|---|
| 5,000 rearranged paths | 5,000 of 5,000 | 4,957 of 5,000 | Guard +43 paths |
| 300 in-order repeated paths | 298 of 300 | 300 of 300 | Always QQQ +2 paths |
Every modeled life began with $750,000 and withdrew $2,500 in the first month. The dollar withdrawal rose with CPI. A life passed only if it funded all 1,200 monthly withdrawals and the inflation-adjusted portfolio trend exceeded 0.1% a year over the 30-, 60-, and 100-year checks. The account-tax calculation reproduces the frozen Swedish ISK-like model used in the original private comparison.
QQQ identity and fund information come from Invesco; SPY identity and fund information come from State Street. The cash return uses the Federal Reserve's three-month Treasury series, and inflation uses U.S. Bureau of Labor Statistics CPI data. Adjusted ETF returns came from licensed historical market data.
These are simulations made by reusing one 25-year U.S. record, not 5,300 independent forecasts. A 100-year path may contain the same historical period more than once. The test does not include every tax system, broker cost, account type, country, or future market environment. It is educational research, not personal investment advice or a live allocation signal.
The next test
Subscribe, then optionally tell us which investment claim or strategy you want checked next.