Why 50+ TAA Strategies All Beat the 4% Rule (And Where Bengen Still Wins)

The 4% rule is the most-cited number in retail retirement planning. William Bengen published it in 1994. SAFEMAX of 4% real for a 60/40 portfolio, 30-year horizon, every single starting year from 1926 to 1976. Withdraw 4% in year one, adjust the dollar amount upward by inflation each year, and you'd never run out across any window in 66 years of US data.

Bengen has since updated the framework substantially in his 2025 book A Richer Retirement and the templates at bengenfs.com. Under the modern formulation, 4% is the recommendation for 50-year FIRE horizons specifically, while traditional 30-year retirement supports closer to 4.7% on a multi-cap rising-glidepath portfolio rather than static 60/40. The 4% folk wisdom is still right, just for a more specific use case than most people quote it. We treat both numbers as relevant benchmarks below depending on the planning horizon.

The other question almost nobody asks: what happens if I don't hold 60/40 (or the rising glidepath) at all? What if I hold a TAA strategy that rotates defensively when the canary flips? What if I hold a capital-efficient stack with built-in diversification?

We computed the rolling-window safe withdrawal rate (SWR) and perpetual withdrawal rate (PWR) for every published strategy in the BestFolio catalog using the same Bengen-style methodology. 51 strategies. 80 variants. Real US CPI inflation from FRED. The results are not subtle: every single published strategy clears Bengen's 4% benchmark, and the best unleveraged TAA strategies sustain withdrawal rates above 14%.

The headline matters less than understanding why, and where the number stops being useful. The rest of this post is the methodology, the table, the mechanism that makes TAA strategies more withdrawal-resilient than static portfolios, and the four honest reasons you should still anchor your retirement plan closer to Bengen's number than to ours.

The methodology

We use the same approach Bengen used in 1994, with three differences: we run it on actual strategy NAVs (not just 60/40), we use FRED's CPI-U for inflation (not Bengen's older inflation series), and we run it programmatically across every strategy variant rather than picking one.

The algorithm:

1. Take the strategy's monthly NAV series since inception (capped at the data window the backtest engine has built up).
2. Choose the rolling-window length: 30 years if the backtest covers at least 30 years, else 20 years, else 10 years.
3. For each rolling N-year window starting at every available month, simulate withdrawing a fixed real dollar amount monthly (inflation-adjusted via FRED CPI for that window).
4. Binary-search for the maximum withdrawal rate where the portfolio survives every single window without going to zero.

That's the SWR. The PWR (perpetual withdrawal rate) is the stricter version: the maximum rate where the portfolio's ending real balance is at least the starting balance in every window. SWR keeps you alive; PWR preserves capital for heirs.

For most strategies in our catalog, the data window starts somewhere between 1986 and 1992 (the earliest common inception date across the underlying assets), giving us 33 to 37 years of NAV history. At a 30-year rolling window, that produces roughly 35 to 84 overlapping windows per strategy. Plenty for a stable estimate, all post-1990.

The "post-1990" detail is the load-bearing caveat. We will come back to it.

The headline result

Sorted by SWR, top to bottom for a representative cut of the catalog. All numbers are real outputs of the BestFolio backtest engine on production data.

The number that should jump out: VAA-G4 SmartStack sustains a 15.3% real withdrawal rate with a maximum drawdown of -19.5%. That's nearly 4x Bengen's 4% rule, with shallower drawdowns than the Classic 60/40 (-34.7%) or the Bogleheads Three-Fund (-36.7%). Same data window, same methodology, completely different outcome.

Even Bengen's own canonical 60/40 portfolio shows up at 6.7% SWR in our methodology, not 4%. That's mostly because we're looking at a more favorable data window (post-1990 has been kind to bonds and gold). Read on for why that matters.

Where Bengen still wins (the honest caveats)

Five reasons you should not actually withdraw 15% from your VAA-G4 SmartStack portfolio.

1. Horizon matters. Our 30-year SWR is too high for 50-year FIRE planning.

The numbers in the table above are rolling 30-year SWRs, computed the same way Bengen's original 1994 paper did. That's the right benchmark if you're planning a traditional retirement starting in your 60s. It's not the right benchmark if you're FIRE planning at 35 with a 50-year horizon ahead.

The relationship between horizon and SWR is non-linear and unfriendly. More years of inflation-adjusted withdrawals means more sequence-of-returns risk to survive, and the SWR drops accordingly. Bengen's own updated work puts the 30-year horizon on a multi-cap rising-glidepath at roughly 4.7% and the 50-year FIRE horizon at the more familiar 4%. Trinity and similar academic updates show 60/40 SWR dropping from around 4% at 30 years to around 3% at 50 years.

For the TAA strategies in our table, expect a similar proportional drop. A 14.2% SWR at 30 years probably becomes something like 11-12% at 50 years. The relative ranking holds (TAA's defensive rotation mechanism gets more valuable over longer horizons, not less), but the absolute numbers come down meaningfully. If you're FIRE planning specifically, mentally derate every SWR in the table above by 20 to 30% before forecasting forward.

2. Our data starts in 1990. Bengen's started in 1926.

This is the most important caveat. Bengen's 4% number came from US stock and bond data starting in 1926, which means his rolling 30-year windows included starting points right before the Great Depression (1929 starts ending in 1959, where the first 10 years lost 70% peak-to-trough), and right before the 1973-1981 stagflation (where US stocks lost real value for nearly a decade while bonds got crushed by rising rates).

Our data windows start somewhere between 1986 and 1992 for most strategies. That window covers the dot-com bust, the GFC, the 2022 inflation shock, and COVID. It does not cover the Depression, the 1937-38 second crash, or the full 1970s stagflation. The 30-year rolling windows we can construct all end before 2022, which means none of them include a sustained inflation regime where bonds AND gold AND TIPS all bled together.

Our SWR numbers are therefore the post-1990 SWR. Adjust expectations downward by 1.5 to 2 percentage points before forecasting forward, because the regime tail risk is real even though our window doesn't see it.

3. SWR is a backtest, not a prediction.

The number "VAA-G4 SmartStack sustained 15.3%" means: in the worst rolling 30-year window of the post-1990 data, that strategy survived a 15.3% real-dollar withdrawal rate. It does not mean future windows will behave the same way. The strategies in our catalog were designed using historical data. Some of their parameters were tuned on the same data the SWR is computed from. There is a non-trivial in-sample-fit component to any SWR computed on a strategy that was published recently.

For strategies in the catalog that have been published for decades (60/40, Permanent Portfolio, GEM, the original BAA, dual momentum), the SWR is closer to an honest out-of-sample number. For the newer SmartStack and SmartLeverage variants (built and published by BestFolio in 2024-2025), the SWR is partially in-sample and should be discounted.

4. SWR doesn't include taxes, fees, or transaction costs.

The backtest assumes monthly rebalancing at end-of-month closing prices with zero transaction cost and zero tax. In reality, monthly rotation in a taxable account generates short-term capital gains every month for a TAA strategy. In a tax-deferred account (IRA, 401k), this is fine. In a taxable account, the after-tax SWR is meaningfully lower.

Rough adjustment: a TAA strategy with monthly turnover in a 24% marginal bracket gives up roughly 50 to 150 basis points of CAGR to tax friction depending on the strategy's turnover rate. That same fraction comes off the SWR.

5. The behavioral SWR is lower than the mathematical SWR.

If you set your retirement plan around 15% SWR and the strategy hits its worst regime in year 3 of retirement (delivering, say, the -34% drawdown of HAA Leveraged 2x), most people change strategy at the bottom. Selling a TAA strategy in defensive mode locks in the loss and forfeits the recovery rule that the strategy was designed around. The math SWR assumes you hold; the behavioral SWR assumes you behave.

Bengen's 4% number quietly accounts for this. 4% is conservative enough that most retirees don't panic-sell when the portfolio is down 30%, because they know they're nowhere near zero. 15% is not conservative enough. The first three years would be terrifying.

Why TAA strategies sustain higher withdrawal rates

The mechanism is sequence-of-returns risk, and it's worth understanding.

A static portfolio in retirement faces a structural problem. If markets crash in years 1-5, the retiree is selling shares at depressed prices to fund spending, locking in losses. The portfolio never gets the recovery benefit because the shares it would have used to recover got spent. Bengen's 4% number is largely about being conservative enough to survive a bad sequence in years 1-5.

TAA strategies attack this directly. When the trend, momentum, or canary signal flips defensive, the strategy is no longer holding the falling asset. HAA's TIP canary, BAA's 4-asset breadth check, GEM's 12-month relative momentum: all of them rotate the portfolio out of equities (or into a less-risky asset) precisely when the bad sequence is starting. The retiree is now spending out of a defensive sleeve that's roughly flat instead of falling.

Two specific cases from the historical record:

2000-2002 dot-com bust. The Classic 60/40 portfolio drew down 24% peak-to-trough over 30 months. A retiree withdrawing 6% per year would have ended that period at roughly 65% of their starting balance, deep in sequence-risk territory. HAA Standard, by contrast, drew down 14% over that period because the TIP canary flipped defensive in November 2000 and stayed defensive through 2001. The same 6% withdrawal in HAA leaves the retiree at 85% of starting balance.

2022 inflation + rates shock. Stocks and long bonds both fell, an unusual correlation that wrecked traditional 60/40 portfolios. Classic 60/40 lost 16% on the year. The TAA strategies with managed-futures sleeves (the SmartStack variants) actually made money in 2022 because the MF overlay was short bonds and short stocks at exactly the right moment. HAA SmartStack returned +6.4% in 2022 while 60/40 was losing 16%. That single year alone explains a chunk of the SWR gap.

The pattern is consistent: TAA strategies don't necessarily earn more in good years (some do, some don't), but they lose less in bad years. Sequence-of-returns risk is asymmetric, so the strategies that protect downside compound the SWR advantage faster than the math suggests.

The unleveraged retirement-grade strategies

If you're using this for actual retirement planning, the leveraged variants are off the table. Holding a 2x leveraged strategy in retirement means a 60% drawdown is on the menu, and the behavioral SWR for a 60% drawdown is approximately zero (you'll sell at the bottom).

The three unleveraged TAA strategies that combine high SWR with bounded drawdowns:

• VAA-G4 SmartStack (Gold+MF). 15.3% SWR / -19.5% MaxDD / Sharpe 1.15. The deepest drawdown in the record was 2022, and even then the SmartStack overlay kept the year roughly flat. Best risk-adjusted profile in the entire catalog.
• HAA SmartStack (Gold+MF). 14.2% SWR / -20.5% MaxDD / Sharpe 1.26. Highest Sharpe in the catalog. TIP canary is a cleaner trigger than VAA's 4-asset breadth in some regimes (faster to flip back risk-on, which captured the 2020 and 2023 rebounds).
• HAA Standard. 12.7% SWR / -19.7% MaxDD / Sharpe 1.19. The vanilla version without the capital-efficient overlay. Easier to implement in a Roth IRA at a US broker (no Avantis-style UCITS gap, no obscure ETFs). The SmartStack adds ~2 percentage points of SWR at the cost of a more complex execution.

Conservative practical recipe for a traditional 30-year retirement: pick one of these as the equity sleeve, hold 10-20% in cash or short-duration Treasuries as a behavioral buffer (you don't have to follow the strategy mechanically in months you don't want to), and target a withdrawal rate around 6 to 7% real. That's still meaningfully above the modern Bengen 4.7% benchmark for 30-year horizons on a multi-cap glidepath, leaves significant margin against the post-1990-window bias, and is psychologically sustainable through a 25-30% drawdown.

For 50-year FIRE planning, derate further. The headline 14% SmartStack number is closer to 11-12% at a 50-year horizon, and the conservative practical target is around 5 to 6% real. Still well above Bengen's 4% updated benchmark, but the cushion is meaningfully narrower than what the 30-year table implies.

If you want to be fully conservative: take a strategy's 30-year SWR, halve it, then round down. HAA SmartStack at 14.2% becomes 7%. VAA-G4 SmartStack at 15.3% becomes 7%. That's still above Bengen's 4% for 50-year FIRE horizons, with cushion for all five caveats above.

What we are not telling you

A short list of additional limitations, in decreasing order of importance.

• Post-1990 bias. Already covered above. Repeated here because it's the biggest issue. Our window misses the worst US regimes for stocks AND bonds simultaneously.
• 30-year rolling windows are overlapping. The 49 windows we get from a 34-year backtest are not 49 independent samples. They share 29 years of overlap. The effective sample size is closer to 5 to 10 truly independent regimes.
• US-only data. Our SWR methodology uses US CPI for inflation. A European or Japanese investor faces different inflation realizations and currency exposure. The SWR for the same strategy held in EUR or JPY is meaningfully different.
• In-sample fit for newer strategies. The SmartStack and SmartLeverage variants were designed in 2024-2025 using data the SWR is now backtested on. There is unavoidable curve-fit risk in any newly-published strategy. The older strategies (60/40, GEM, BAA, HAA Standard, ADM, VAA) are out-of-sample for the data window we're testing on, which is genuinely useful.
• Bengen used SAFEMAX with explicit failure on the worst window. So do we. But we use binary search to find the rate; in extreme cases the binary search can converge to a number that's exactly the worst-window rate with no margin. Add 50 to 100 basis points of safety margin to any cited SWR.
• CAPE valuation adjustment. Bengen's updated framework includes a CAPE-based adjustment: when starting valuations are high (as they are in 2026), the safe withdrawal rate should be derated. Our analysis does not account for this. The implied derate at current CAPE levels is in the 50-100 basis point range relative to a long-run average starting CAPE.
• Single-asset failure modes. If a TAA strategy holds 100% TLT in a defensive month and TLT loses 7% that month (like several months in 2022), the strategy bleeds. The diversified SmartStack variants are more robust to this than the binary canary strategies because the defensive position still has a managed-futures overlay rather than concentrating in one bond ETF.

Practical takeaway

The 4% rule is conservative for almost every strategy in the BestFolio catalog. That's the actual finding. The 4% rule is also probably the right number to use anyway, for behavioral reasons and for regime-tail-risk reasons that our 33-year data window can't see.

The genuinely useful takeaway is that the gap between strategies is enormous. A retiree holding Permanent Portfolio (5.8% SWR) versus HAA SmartStack (14.2% SWR) has very different headroom against bad sequences. They both clear the Bengen number, but one has 200% more cushion than the other for the same retirement horizon. That matters whether you withdraw 4% or 6% or 8%.

If you want to see the SWR for any specific strategy in your portfolio, the BestFolio strategy detail pages now show it alongside CAGR, MaxDD, and Sharpe. Browse the catalog or build a custom portfolio and the engine computes the rolling-window SWR on the blended NAV directly.

All numbers in this post come from BestFolio's backtest engine on production data, computed via rolling-window Bengen methodology with real US CPI from the FRED API. Data window 1986-2026 depending on strategy. The methodology source is at backend/app/services/withdrawal_rate.py if you want to inspect the algorithm directly.