Data through October 2026. Monthly rebalance, trading costs included.
How Defense First works
Most tactical strategies own stocks by default and step aside when a risk signal fires. Defense First starts from the other end. Each month it ranks four defensive assets, each tied to a different kind of market stress, by the plain average of their 1, 3, 6 and 12-month returns. It holds them at 40%, 30%, 20% and 10% by rank.
| Role | ETF | Hedges against |
|---|
| Long-term Treasuries | TLT | Deflation, disinflation, Fed easing |
| Gold | SGOL | Monetary instability, falling real rates |
| Commodities | PDBC | Inflation, supply shocks |
| US dollar | UUP | Global funding stress, flight to safety |
| Equity fallback | VOO | Takes any slot whose defensive asset is losing to T-bills |
The one rule that gives it its name: a defensive asset whose momentum score is below Treasury bills gives up its slot, and that weight goes to the S&P 500. When all four defensive assets are strong it holds no stocks. When they all fade it is fully invested in stocks. Over this backtest its stock allocation has averaged about a third. The portfolio is rebalanced every month.
The rules come from Carlson's 2025 paper, Defense First: A Multi-Asset Tactical Model for Adaptive Downside Protection.
The leveraged revision
In October 2026 Carlson published a revision, called DF New in his comparison and run here as Defense First Engineered. Only the stock fallback changes. Half of it goes to a 2x Nasdaq-100 fund (QLD) at all times. The other half holds an anti-beta fund (BTAL, long low-beta stocks and short high-beta stocks) when BTAL is beating T-bills, and QLD otherwise. The defensive assets and ranking are untouched.
His results covered January 2013 onward, which begins shortly after BTAL launched. This backtest runs both versions back to 1980, so the revision faces the 2000 to 2002 bear market and the 2008 financial crisis, not just the long Nasdaq-led bull market that followed.
What the record shows
From Jan 1980 through Oct 2026, Defense First compounded at 10.8% a year after trading costs, turning $10,000 into about $1,225,132. The S&P 500 returned 12.0% a year over the same months, with a maximum drawdown of -51.0% against -16.8% for Defense First.
Its deepest decline was 16.8%, from a peak in Jan 1980 to a low in Mar 1980, and it was back to its old high by Sep 1980. That decline came from its defensive assets, when gold and commodities reversed from their 1980 highs, not from a stock market crash.
The crises are where the design shows. Through the dot-com bear market Defense First returned -5.8% while the S&P 500 returned -44.8%. Through the global financial crisis it returned +5.6% against -51.0%. In 2022, when stocks and bonds fell together, it returned +7.8% against -18.2%.
The leveraged revision earned more, 13.3% a year against 10.8%, but its maximum drawdown was -42.3% against -16.8%. Nearly all of that came in the dot-com bear market, when it returned -39.4%, because nothing in the revision protects its 2x Nasdaq-100 core when the Nasdaq-100 itself is collapsing. On the Ulcer Performance Index, which weighs return against the depth and length of drawdowns, Defense First scores 1.66 and the revision 0.76.
Among the strategies on this page, Triad, a Dual Momentum Systems strategy, returned 14.4% a year with a -13.3% maximum drawdown over the same months, for an Ulcer Performance Index of 3.67 against 1.66 for Defense First.
In the 196 months the S&P 500 fell, Defense First still made money in 36% of them.
Since 2000
The modern half of the record, with the dot-com bust, 2008, the long zero-rate decade and the 2022 rate shock all inside it.
| Jan 2000 to Oct 2026 | CAGR | Max DD | Ulcer | UPI | Sortino | MAR |
|---|
| Defense First | 10.1% | -15.0% | 4.16 | 1.99 | 2.12 | 0.68 |
|---|
| Defense First Engineered | 12.8% | -42.3% | 12.66 | 0.87 | 1.93 | 0.30 |
|---|
| Triad | 14.2% | -8.7% | 2.58 | 4.77 | 2.90 | 1.63 |
|---|
| S&P 500 | 8.0% | -51.0% | 15.87 | 0.39 | 0.88 | 0.16 |
|---|
| HAA - Hybrid Asset Allocation | 11.8% | -9.4% | 2.52 | 3.95 | 2.50 | 1.25 |
|---|
| Global Navigator | 13.0% | -16.1% | 4.65 | 2.39 | 2.06 | 0.81 |
|---|
| 60/40 | 6.6% | -32.3% | 7.61 | 0.62 | 1.11 | 0.20 |
|---|
Return by decade
| Annual return | 1980s | 1990s | 2000s | 2010s | 2020s* |
|---|
| Defense First | +12.4% | +11.0% | +9.4% | +6.3% | +17.2% |
|---|
| Defense First Engineered | +9.1% | +18.8% | +6.5% | +12.3% | +23.7% |
|---|
| Triad | +18.3% | +11.1% | +12.8% | +11.7% | +20.0% |
|---|
| S&P 500 | +17.0% | +18.1% | -1.0% | +12.9% | +14.9% |
|---|
| HAA - Hybrid Asset Allocation | +21.9% | +13.3% | +14.5% | +8.3% | +13.2% |
|---|
| Global Navigator | +17.3% | +14.3% | +13.2% | +12.4% | +13.6% |
|---|
| 60/40 | +15.5% | +14.0% | +2.1% | +9.4% | +9.3% |
|---|
* Partial decade, annualized over the months available.
Through the hard stretches
Total return over each period, start of the first month to the end of the last.
| Period | Defense First | DF Engineered | Triad | S&P 500 | HAA | Global Navigator | 60/40 |
|---|
| 1987 crashSep 1987 to Nov 1987 | -4.9% | -14.4% | -6.2% | -29.8% | +1.4% | -17.7% | -17.7% |
|---|
| Dot-com bear marketSep 2000 to Sep 2002 | -5.8% | -39.4% | +3.9% | -44.8% | +9.3% | +12.7% | -23.5% |
|---|
| Global financial crisisNov 2007 to Feb 2009 | +5.6% | -0.1% | +7.6% | -51.0% | -1.3% | -8.4% | -32.3% |
|---|
| COVID crashFeb 2020 to Mar 2020 | +3.3% | +4.2% | -0.1% | -19.6% | +2.9% | -0.8% | -11.9% |
|---|
| 2022 rate shockJan 2022 to Dec 2022 | +7.8% | +4.0% | +1.7% | -18.2% | +3.1% | -16.1% | -15.9% |
|---|
How it behaves month to month
| Since Jan 1980 | Worst 12 months | Positive months | Positive when S&P 500 fell | Correlation to S&P 500 |
|---|
| Defense First | -15.0% | 63% | 36% | 0.48 |
|---|
| Defense First Engineered | -35.7% | 61% | 34% | 0.52 |
|---|
| Triad | -9.7% | 70% | 40% | 0.49 |
|---|
| HAA - Hybrid Asset Allocation | -8.7% | 71% | 45% | 0.46 |
|---|
| Global Navigator | -21.8% | 70% | 34% | 0.54 |
|---|
| 60/40 | -27.5% | 67% | 10% | 0.98 |
|---|
The strategies it is compared with
- Defense First Engineered: Tom Carlson's October 2026 revision ("DF New"). The same defensive rules, with the stock fallback replaced by 2x Nasdaq-100 and an anti-beta fund.
- Triad: Dual Momentum Systems. Three independent momentum sleeves (real assets, US large caps, and a relative-strength sleeve), each stepping to intermediate Treasuries when its signal turns.
- S&P 500: US large-cap stocks, held without rebalancing.
- HAA - Hybrid Asset Allocation: Keller and Keuning, 2023. One TIPS canary decides between eight global assets and Treasuries.
- Global Navigator: Dual Momentum Systems. Dual momentum between US and international equities, stepping out of equities when neither beats cash.
- 60/40: 60% US stocks, 40% US bonds, rebalanced annually.
Questions
What is the Defense First strategy?
Defense First is a tactical asset allocation strategy that ranks four defensive assets each month (long-term Treasuries, gold, commodities and the US dollar) and holds them at 40%, 30%, 20% and 10% by momentum rank. Any defensive asset whose momentum is below Treasury bills hands its slot to the S&P 500, so stock exposure is whatever the defensive assets leave behind.
Who created Defense First?
Tom Carlson, a former actuary and retirement executive, who published it in 2025 in the paper "Defense First: A Multi-Asset Tactical Model for Adaptive Downside Protection" and placed it in the public domain. Dual Momentum Systems is not affiliated with him. This page is an independent backtest.
What has Defense First returned?
In this backtest, from Jan 1980 through Oct 2026, it returned 10.8% a year after trading costs, with a maximum drawdown of -16.8%. The S&P 500 returned 12.0% a year with a maximum drawdown of -51.0%. The figures update every month.
Is the leveraged Defense First revision (DF New) better?
It has earned more, 13.3% a year against 10.8% over the same months, but with a maximum drawdown of -42.3% against -16.8%, most of it in the 2000 to 2002 bear market. Its results since 2013 look far stronger than its full history, because that period was unusually kind to the Nasdaq-100.
How often is Defense First rebalanced?
Monthly. Signals use month-end data, and the whole portfolio is reset to its target weights each month whether or not the ranking changed.
Which ETFs does this backtest use?
TLT for long-term Treasuries, SGOL for gold, PDBC for commodities, UUP for the US dollar and VOO for the stock fallback, with BIL as the T-bill hurdle. These are DMS's choice of fund for each role.
Why these numbers differ from the paper
- Longer history. This backtest starts in 1980 and uses each fund's index history before launch. The US dollar fund (UUP) is rebuilt from the US Dollar Index's currency weights plus the interest-rate carry the fund earns. For the revision, QLD is 2x the daily Nasdaq-100, and BTAL is a low-minus-high beta portfolio from the Ken French data library, calibrated to BTAL's live returns.
- Fund choices. Gold, commodities and stocks use this site's standard funds (SGOL, PDBC, VOO) in place of the paper's GLD, DBC and SPY. The exposures are the same.
- Trading costs included. Every figure is after a per-trade cost that varies with each fund's spread and with market stress. See where the data comes from.
About the data
All portfolios run on the same monthly return series with the same trading-cost model and are rebalanced as each one specifies. Returns before an ETF's launch come from a matching index, as explained in where the data comes from. Before each fund launched, its history comes from a matching index or a documented reconstruction. The anti-beta fund used by Defense First Engineered has the loosest reconstruction of any series here; it matches the live fund's volatility and market sensitivity closely but not its month-to-month path, so treat the revision's pre-2011 detail with some caution.
Backtests are hypothetical and are not a promise of future results. This page is educational and is not investment advice.