Strategy in short: Hold QQQ if SPY beat T-bills in at least five of the past 12 completed months. Otherwise, hold T-bills.
Retirement answer: The guard supported more starting spending: 8.6% versus 5.9% a year funded 30 years in at least 95% of both sets of histories. These unusually high figures reuse one strong 25-year market record and are not recommended spending.
Return comparison: In this 25-year backtest, $10,000 grew to $374,082 with the guard and $190,801 while always holding QQQ.
This is promising because the guard protected capital during parts of the 2001–02 technology decline and the 2008–09 financial crisis, then returned a larger balance to QQQ. It backfired in early 2003, missing a 12.6% QQQ rebound, but the losses it avoided were much larger. The deepest fall was about 33% with the guard, compared with about 55% while always holding QQQ.
See when the guard helped—and when it hurt
The strategy moved out of QQQ for 22 months. During those months, QQQ fell 48% while T-bills gained 2.7%. This is where the extra return came from.
QQQ fell in 13 of those 22 months. Those down months compounded to a 72% loss. QQQ rose in the other 9 months, compounding to an 86% gain. The long falling stretches outweighed the rebound the strategy missed.
What was happening?
2001–03 · Technology bust
The excitement around the internet had outrun company results.
Investors had poured money into technology companies and infrastructure. When earnings slowed and excess capacity became obvious, financing dried up and many companies failed. Technology shares kept falling through several attempted recoveries. The San Francisco Fed described the cycle as it unfolded.
What it felt like: QQQ repeatedly looked ready to recover, then fell to a new low. The rule moved to T-bills three times as the weakness persisted, then missed the first 2003 rebound.
2008–09 · Financial crisis
A housing crash spread through banks, lending and jobs.
Losses on mortgages damaged financial institutions and made credit harder to obtain. By autumn 2008, failures and frozen markets had turned a housing downturn into a deep recession. The Federal Reserve's history explains how the crisis spread.
What it felt like: Investments, home values and the job market were all under pressure. The rule reacted in October 2008 and stayed in T-bills through March 2009, while QQQ lost 21.8%.
Could this happen again?
Yes. Another long decline is a real risk during a 30-year FIRE life.
Two episodes cannot give us an honest probability. They do show the kind of market this rule helped with: broad weakness that lasted long enough to repeat across months. The rule stayed in QQQ through the fast 2020 crash and the 2022 decline.
Today has both a similarity and an important difference. Technology made up 59.8% of the Nasdaq-100 on 31 March 2026, so disappointment across large technology companies could still hit the index hard. But all 32 large banks in the Federal Reserve's 2026 stress test stayed above their minimum capital requirements during a severe hypothetical recession. The next long decline could come through a different route than either 2001 or 2008.
The guard beat QQQThe guard lost to QQQ
Growth of $10,000 — logarithmic scale
This graph uses a logarithmic scale: the same percentage change takes up the same vertical space. Each shaded band is a period spent in T-bills. Its label shows how much the guard gained or lost against QQQ during that exit. Eight extra switches reduced the final advantage by about 0.8%.
What happened each time the guard held T-bills
What happened each time the guard held T-bills
Time in T-bills
T-bills
QQQ
Effect
July–December 2001
+1.3%
-14.9%
Protected
February–March 2002
+0.3%
-6.4%
Protected
May–August 2002
+0.6%
-26.0%
Protected
January–April 2003
+0.4%
+12.6%
Missed rebound
October 2008–March 2009
+0.1%
-21.8%
Protected
The table counts only months spent in T-bills. Both portfolios held QQQ during every other month.
What if we don't have a new IT bubble?
Growth of $10,000 after the technology bust — logarithmic scale
The guard beat QQQ
This starts a fresh $10,000 portfolio in May 2003, after the technology-bust-era exits ended. The opening decision uses the twelve months that had already finished. The shaded band is the only later period spent in T-bills.
The guard still helped. $10,000 grew to $406,074 with the guard and $317,654 holding QQQ. The deepest fall was 32.6% with the guard and 49.7% holding QQQ.
The two portfolios differed for only 6 months: October 2008 through March 2009. QQQ fell 21.8% while T-bills gained 0.1%. Avoiding part of that financial-crisis decline left the guard with about 28% more at the end.
The result does not depend only on another technology bubble. It does depend on one other long crisis, so this later test is useful evidence—not proof that the guard will help in every future decline.
Guarded QQQ15.6% a year-32.6% deepest fall
Always QQQ12.5% a year-54.7% deepest fall
July 2001–June 2026. This is a hypothetical backtest, not an account someone actually held. Past results do not predict future results. Untick either line to view the other without changing the scale.
Can the guard support taking out money every month?
Yes. The guard funded more retirements in both history tests. It protected capital during long declines, so monthly withdrawals sold less of the portfolio and more remained for the recovery.
The guard helped when it moved to T-bills during long declines. A smaller fall left more capital to return to QQQ while monthly withdrawals continued. Missing a rebound had the opposite effect.
Example: You have $500,000 and need $2,500 a month—6% a year.
Before either fall$500,000$2,500 a month starts at 6.0% a year.
After the guard's deepest fall$337,107 left
8.9% a yearThe same spending
After QQQ's deepest fall$226,469 left
13.2% a yearThe same spending
Each line applies that strategy's own deepest tested fall to the same $500,000 starting portfolio, before counting any withdrawals. The longer line means more capital remained for the recovery.
The guard's smaller deepest fall left about $110,638 more in this illustration. That larger base made the same monthly spending less damaging while QQQ recovered.
How much spending survived at least 95% of both history tests?
For a 30-year retirement, the highest starting withdrawal rate was 8.6% with the guard and 5.9% while always holding QQQ. On a $500,000 portfolio, the guard supported $1,125 more starting spending each month in these tests.
Guarded QQQ · 8.6% a year$3,583 a month295 of 300 ordered starts · 4,775 of 5,000 mixed historiesAlways QQQ · 5.9% a year$2,458 a month288 of 300 ordered starts · 4,823 of 5,000 mixed histories
At the same 5.9% starting rate, the guard funded 300 of 300 ordered starts and 5,000 of 5,000 mixed histories. Always holding QQQ funded 288 and 4,823. The guard was the stronger withdrawal strategy in both tests.
The 8.6% figure is unusually high because every test reuses the same strong 25-year market record. It is a historical threshold, not a safe withdrawal recommendation or a promise of future success.
Test whether SPY-guarded QQQ can fund your retirement with your own numbers
Enter your portfolio, monthly spending, and how long the money needs to last.
$2,500 a month for 30 years starts at 6.0% of the portfolio each year.
In these backtests, almost always. Both strategies funded this FIRE life in at least 9 out of 10 histories tested.
300 starting months; the fund's record repeats in order when needed
Guarded QQQ
300 funded
Always QQQ
284 funded
5,000 histories built from 8–12-year pieces
Guarded QQQ
5,000 funded
Always QQQ
4,804 funded
A path funded the goal only if it paid every inflation-adjusted monthly withdrawal for 30 years. Results use the nearest 0.1 percentage-point starting withdrawal rate. They reuse one 25-year market record and are not a probability about your future.
How the FIRE calculator works
The calculator applies the portfolio value, monthly withdrawal, and number of years you enter. The first withdrawal occurs before the first month's return, and later withdrawals rise with U.S. inflation. A history counts as funded only if it pays every monthly withdrawal for the full period and finishes above zero.
One result begins at each of the 300 months in the fund record and follows that record forward in order, returning to its first month when more months are needed. The other builds 5,000 longer histories from linked 8-, 10-, and 12-year pieces of the same record. The calculation charges 0.10% whenever QQQ is bought or sold and excludes personal tax.
When a simulated history jumps to another historical period, the guard uses the 12 real months that preceded the incoming month. The retirement account and its withdrawals continue without resetting.
Strategy in detail
At the end of every month, the rule compares SPY with cash during each of the previous twelve completed months. If SPY beat cash in at least five of them, the portfolio holds QQQ for the next month. Otherwise, it holds cash earning the three-month Treasury rate.
The rule makes a new choice every month. It moves into T-bills after SPY has failed to beat them in at least eight of the previous 12 months.
Research supports the idea that market moves can continue for months. The 2012 Time Series Momentum study found that recent direction tended to persist for one to 12 months across 58 equity-index, bond, currency, and commodity futures. A later study of 67 markets back to 1880 found that simple trend following remained profitable on average across very different periods. That gives this rule a plausible reason to leave QQQ after broad weakness has persisted.
In my tests, the signal most clearly identified slow declines that persisted for months. Those were the periods when a monthly rule had time to move into T-bills. Switching cost little in most other periods.
Can I trust these results?
My evidence-weighted estimate is about a 65% chance that this exact rule would improve a long retirement compared with always holding QQQ.
Here, success means funding at least as many planned withdrawals while reducing the deepest fall. The 60% is my judgment after weighing the evidence, not a measured probability or a promise about the future.
The retirement evidence favors the guard. It funded more of the ordered starting months and the mixed long histories, while its smaller deepest fall left more capital available for the recovery. That raises confidence in the rule as retirement protection.
The 65% estimate also reflects that the guard grew more, fell less, used only information available at the time, and stayed ahead after distributions and trading costs. Confidence remains limited because every public test reuses the same 25 years. Rearranging those years creates different retirement paths, not new market history. The result excludes personal tax and compares the rule with holding QQQ rather than every sensible retirement portfolio.
Between these two choices, the evidence favors the guard. It is not strong enough to treat the rule as proven or assume its historical withdrawal threshold is safe.
Methodology details for the nerds
How I ran the return comparison
The graph starts with $10,000 in July 2001 and follows the 300 real months through June 2026 in their original order. The comparison holds QQQ throughout. The guarded portfolio holds QQQ for 278 months and Treasury cash for 22 months. Distributions are reinvested. I subtract 0.10% whenever QQQ is bought or sold. The graph includes no withdrawals, inflation adjustment, or personal tax.
A new portfolio begins in July 2001, but its first decision uses the twelve completed months from July 2000 through June 2001. Those earlier months supply the signal only; they add no return to the portfolio. Every later decision follows the same rule.
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.
FINRA's investor guide to backtested strategies explains the central limit: historical simulations can help compare rules, but they cannot tell us how a strategy will perform in the future or fully reproduce real investing.
See how much you need to retire—and compare what to invest in.
Enter what you have and what you want to spend each month. Compare three portfolios against the same market history. Your financial figures stay in your browser.
