Defensive Asset Allocation (DAA) Overview
| CAGR | 11.3% |
|---|---|
| Maximum drawdown | -20.2% |
| Ulcer Index | 5.09 |
| UPI | 1.40 |
| History | 46 years (560 months) |
Defensive Asset Allocation (DAA) is Wouter Keller and Jan Willem Keuning's 2018 strategy, and the one that introduced the idea every later model in their line depends on: the canary universe.
It watches two assets that it never holds, emerging market equities and aggregate bonds. Both healthy, and the portfolio is fully invested in the six strongest of twelve global assets. One of the two failing, and half the portfolio moves to bonds. Both failing, and all of it does.
Dual Momentum Systems tracks DAA as a benchmark.
The Backstory
The problem with watching what you own
Their previous model, Vigilant Asset Allocation, used the same set of assets for two different jobs: deciding what to buy, and deciding whether to be invested at all. Most tactical strategies do this, because it seems natural. If the things you might buy look bad, hold cash instead.
The trouble is that the two jobs want different things from a universe. Selection wants a broad list of assets you would be happy to own. Detection wants assets that react early and reliably to trouble, whether or not you would ever hold them. A list built for one job will be mediocre at the other.
The practical symptom was that Vigilant sat in cash roughly 60% of the time. It was well protected, and it was out of the market so often that it gave up a great deal of return to get there.
The canary universe
DAA's answer was to split the jobs apart. Keep the twelve-asset list for choosing what to hold, and give the protection decision its own separate two-asset universe whose only job is to sound an alarm.
The name comes from the canary in the coal mine. These assets are not held, not ranked, not part of the portfolio in any way. They exist to be sensitive.
The pair they settled on was emerging market equities and aggregate bonds. The logic is that the two cover different kinds of trouble. Emerging markets are the high-beta end of global equities and tend to break first when risk appetite fades. Aggregate bonds respond to interest rates and credit conditions, which is a different failure mode entirely. Trouble that shows up in neither is usually not the kind that requires hiding.
The result was exactly what they were aiming for. DAA's average time spent defensive is roughly half of Vigilant's, while the crash protection stays nearly as good. Less time hiding, similar safety.
How they chose the canaries, which is the interesting part
The obvious objection to a two-asset alarm is that you could find some pair that would have worked beautifully on any given history. Pick through enough combinations and something always fits.
Their method addressed that directly. They searched for the canary pair using a deliberately crude model over 1926 to 1970, holding only the S&P 500 as the risky asset, then applied the winning pair to the real strategy over 1970 to 2018. The period that chose the alarm and the period that tested the strategy do not overlap at all.
That is a more careful piece of work than most published tactical strategies bother with, and it is a large part of why the canary idea was taken seriously enough to become standard in everything they published afterward. Bold Asset Allocation uses four canaries. Hybrid Asset Allocation uses one. Both trace back to this paper.
Why this site carries it
DAA is here as a benchmark and as the origin point of a family of strategies. Reading it alongside Vigilant shows what separating the alarm from the shopping list actually bought, and reading it alongside Hybrid shows where the idea ended up five years later. Running all of them on the same data with the same trading costs is the only way to make those comparisons honestly.
Core Strategy Logic
DAA is evaluated once a month. Signals come from month-end data and set the holdings for the following month.
Everything is scored with the same weighted momentum measure, which averages 1, 3, 6 and 12-month returns with the most recent month weighted most heavily. It is a fast measure, chosen so the alarm reacts promptly.
Step 1 - Count the failing canaries
Score emerging market equities (EEM) and aggregate bonds (BND). Count how many have negative momentum. That count, zero, one or two, decides everything about how the portfolio is positioned.
Step 2 - Rank the risky universe and pick the defensive bond
Score all twelve risky assets and rank them. Score the three defensive candidates and take the single best.
Step 3 - Allocate according to the count
| Failing canaries | Allocation |
|---|---|
| None | 100% split equally across the top six risky assets |
| One | 50% split equally across the top three risky assets, 50% in the best defensive bond |
| Both | 100% in the best defensive bond |
Note what happens in the middle case. The portfolio does not merely hold less risk, it holds more concentrated risk: the top three rather than the top six. Half the money comes off the table and the half that stays goes into the strongest names only. That is a sharper response than a simple proportional reduction.
The portfolio is rebalanced to fresh targets every month.
The universes
Canary, signal only, never held on this basis:
| Asset | What it detects |
|---|---|
| EEM | Risk appetite in global equities |
| BND | Interest rate and credit conditions |
Risky, twelve candidates, top six or top three held:
| Category | Assets |
|---|---|
| U.S. equities | large cap (VOO), Nasdaq 100 (QQQ), small cap (IWM) |
| International equities | Europe (VGK), Japan (EWJ), emerging markets (EEM) |
| Real assets | real estate (VNQ), commodities (PDBC), gold (SGOL) |
| Credit and duration | long Treasuries (TLT), high yield (HYG), corporate bonds (LQD) |
Defensive, three candidates, best one held:
| Asset | Role |
|---|---|
| SHY | Short Treasuries |
| IEF | Intermediate Treasuries |
| LQD | Corporate bonds |
Two things are worth noticing. Emerging market equities appear as a canary and as a risky candidate, so the same asset can both raise the alarm and be held. Corporate bonds appear in both the risky and defensive lists. Neither is an error; both are in the published design.
The defensive universe is also worth a second look, because it is all bonds and no cash equivalent. In a year like 2022, when both stocks and bonds fell, DAA's safe harbor was itself losing money. Later models in this line, particularly Bold Asset Allocation, widened the defensive universe specifically to address that.
Performance Highlights
Over the full published history, alongside the model it improved on and the model that eventually replaced the idea:
January 1980 through August 2026, net of trading friction:
| CAGR | Max Drawdown | UPI | |
|---|---|---|---|
| Defensive Asset Allocation | +11.3% | -20.2% | 1.40 |
| Vigilant Asset Allocation - Aggressive | +12.4% | -21.6% | 1.26 |
| Hybrid Asset Allocation | +14.1% | -9.6% | 3.64 |
| Triad | +14.9% | -13.8% | 3.94 |
| S&P 500 | +12.0% | -51.0% | 0.61 |
The third column is the Ulcer Performance Index: return above cash divided by how deep the strategy went and how long it stayed there. Max Drawdown reports the worst single moment; UPI reports the experience of holding it. Higher is better.
Against Vigilant, the model it was built to improve, DAA takes a slightly smaller worst decline and scores better on UPI while giving up some annual return. The canary idea did what the paper claimed on the risk side. The return improvement the authors reported is less visible in this implementation over this period, which is a useful reminder that a published result and an independent reproduction rarely land in exactly the same place.
Against the index, DAA earns a little less per year while holding its worst decline to about 40% of the index's, and its UPI is more than double.
The comparison that matters most is with Hybrid Asset Allocation, five years further down the same line of thinking. HAA earns considerably more, declines half as much at its worst and more than doubles the UPI. The canary concept was right; DAA was an early version of it.
The same five, measured from January 2000:
January 2000 through August 2026, net of trading friction:
| CAGR | Max Drawdown | UPI | |
|---|---|---|---|
| Defensive Asset Allocation | +8.3% | -20.2% | 1.04 |
| Vigilant Asset Allocation - Aggressive | +10.2% | -21.6% | 1.09 |
| Hybrid Asset Allocation | +11.5% | -9.6% | 3.56 |
| Triad | +14.2% | -8.6% | 4.93 |
| S&P 500 | +8.0% | -51.0% | 0.39 |
The recent window is harder on DAA. It barely beats the index on return, though still with a far smaller decline, and it now trails Vigilant on both return and UPI. Its worst decline is identical in both tables, meaning the deepest drawdown of the full record happened after 2000, and an all-bond defensive universe during 2022 is the likely explanation.
Two cautions apply as always. A window starting in January 2000 begins shortly before a major bear market, which flatters strategies that step aside during one. And a single figure covering decades says nothing about the order the returns arrived in, which is most of what an investor lives through.
Portfolio Characteristics
- Separate alarm and shopping list. The first strategy to detect trouble with assets it does not hold.
- Three positions, not a dial. Fully invested, half invested, or fully defensive. Nothing in between.
- Concentration increases as risk comes off. The half-invested state holds the top three rather than the top six.
- Less time hiding than its predecessor. Roughly half Vigilant's time spent defensive, with comparable protection.
- All-bond defense. The defensive universe holds no cash equivalent, so a year when bonds fall is a year the safe harbor leaks.
- Fast momentum measure. Weighted toward the most recent month, so the alarm reacts quickly and can also whipsaw.
- Broad risky universe. Twelve candidates spanning equities, real assets, credit and duration.
- Monthly rebalancing. Fresh targets every month.
- No leverage. DAA never holds a leveraged fund.
- Best suited to tax-deferred accounts. Monthly rebalancing and frequent state changes generate short-term gains. In a taxable account, plan accordingly.
- Fully mechanical. Every allocation follows from published rules with no discretionary override.
Who It's For
DAA is designed for investors who want:
- The original canary-based design, run as published.
- Protection that is decided separately from selection.
- A middle gear between fully invested and fully defensive.
- Broad diversification across twelve candidate assets when conditions are calm.
- A rules-based process with no market-timing judgment calls.
It is a poor fit for investors who need shallow drawdowns, who are uneasy about a defensive position made entirely of bonds, or who trade in a taxable account. Investors drawn to the canary concept should look at Hybrid Asset Allocation, which carries the idea forward with better results, or Bold Asset Allocation, which pairs it with a defensive universe that can hold real assets. Those wanting a comparable approach from strategies built here should look at Triad.
Defensive Asset Allocation was created by Wouter Keller and Jan Willem Keuning and is presented here as an independent implementation, with standard fund substitutions. Dual Momentum Systems is not affiliated with or endorsed by either author. The original paper is available at SSRN.
Past performance, including backtested results, is not indicative of future results. Backtested performance is hypothetical, does not reflect actual trading, and benefits from hindsight in the selection of strategy rules. Portions of the long-term history rely on reconstructed proxy data for funds that did not exist for the full period. Investors should carefully consider their risk tolerance and consult with a financial advisor.
For the latest details, visit www.DualMomentumSystems.com