Vigilant Asset Allocation G12 (VAA-G12)
Vigilant Asset Allocation across twelve global asset classes. It counts how many of the twelve have negative weighted momentum: with none or one, it holds the two strongest; with two or three, half moves to the strongest cash-like bond fund; with four or more, it holds only that bond fund.
Designed by Wouter Keller and JW Keuning, 2017. Implemented and tracked by Tactfolio.
| Apr 2008 – Sep 2026 | Strategy | SPY |
|---|---|---|
| Annual return (CAGR) | 7.2% | 11.9% |
| Worst drawdown | -27.4% | -51.5% |
| Sharpe ratio | 0.64 | 0.67 |
| Volatility | 12.1% | 19.7% |
| Annual return since publication (Aug 2017) | 6.6% | 15.0% |
| Sep 1987 – Sep 2026 | Strategy | SPY |
|---|---|---|
| Annual return (CAGR) | 9.9% | 10.7% |
| Worst drawdown | -27.4% | -55.2% |
| Sharpe ratio | 0.87 | 0.64 |
| Volatility | 11.5% | 18.5% |
| Annual return since publication (Aug 2017) | 6.6% | 15.0% |
VAA-G12 is the twelve-asset version of Vigilant Asset Allocation, published in 2017 by Wouter Keller and JW Keuning. It applies the same breadth momentum idea as the better-known VAA-G4 to a broad global universe, and turns defensive in stages rather than all at once.
The idea#
Keller and Keuning's breadth momentum treats bad momentum anywhere in the universe as a warning. With twelve asset classes, one or two weak assets are normal, so VAA-G12 counts how many are weak and scales into safety as the count rises. The portfolio holds only the two strongest assets while conditions are good, which keeps it concentrated in whatever is leading.
How it works#
At the close of the last trading day of each month, score twelve funds by Keller's 13612W momentum (a weighted average of the 1-, 3-, 6-, and 12-month returns, with weights 12, 4, 2, and 1): SPY, IWM, QQQ, VGK, EWJ, EEM, VNQ, GSG, GLD, TLT, HYG, and LQD. Then count the funds with zero or negative momentum:
- None or one: hold the two strongest funds, half each.
- Two or three: hold the strongest fund with half the portfolio, and the strongest of short Treasuries (SHY), intermediate Treasuries (IEF), and corporate bonds (LQD) with the other half.
- Four or more: hold only that strongest bond fund.
What the backtest shows#
Over the ETF era, from 2008, VAA-G12 made about 7% a year. It trailed both the S&P 500 and a 60/40 portfolio on risk-adjusted return. Its record is uneven: it gained more than 20% in 2008 by moving to bonds early, and again in 2020, but it lost in the 2009 rebound and trailed badly in 2023 and 2024.
Its worst drawdown, about 27%, came in 2022 and 2023, when rising rates hurt both its stock picks and its bond fallback. It took until 2025 to recover.
With simulated history the test starts in 1987 and shows a higher return, near 10% a year, with the same worst drawdown.
When it struggles#
- Broad bond sell-offs. Its defensive funds are all bonds, so rising rates hurt it on both sides.
- Two-asset concentration. Holding only two funds means a wrong pick weighs heavily.
- Rebounds. It can stay partly defensive while a recovery is under way.
Using it on Tactfolio#
The live strategy above implements the counting rule exactly. Copy it to change how many funds it holds or where the defensive steps start.
Year by year
| Year | Strategy | SPY |
|---|---|---|
| 2026* | -1.5% | 14.0% |
| 2025 | 24.9% | 17.7% |
| 2024 | 4.2% | 24.9% |
| 2023 | 6.5% | 26.2% |
| 2022 | -14.2% | -18.2% |
| 2021 | 2.0% | 28.7% |
| 2020 | 23.5% | 18.3% |
| 2019 | 7.8% | 31.2% |
| 2018 | 10.0% | -4.6% |
| 2017 | 14.5% | 21.7% |
| 2016 | 3.4% | 12.0% |
| 2015 | -0.4% | 1.2% |
| 2014 | 5.9% | 13.5% |
| 2013 | 1.9% | 32.3% |
| 2012 | 10.5% | 16.0% |
| 2011 | 11.9% | 1.9% |
| 2010 | 12.9% | 15.1% |
| 2009 | -3.8% | 26.4% |
| 2008* | 21.7% | -32.4% |
| Year | Strategy | SPY |
|---|---|---|
| 2026* | -1.5% | 14.0% |
| 2025 | 24.9% | 17.7% |
| 2024 | 4.2% | 24.9% |
| 2023 | 6.5% | 26.2% |
| 2022 | -14.2% | -18.2% |
| 2021 | 2.0% | 28.7% |
| 2020 | 23.5% | 18.3% |
| 2019 | 7.8% | 31.2% |
| 2018 | 10.0% | -4.6% |
| 2017 | 14.5% | 21.7% |
| 2016 | 3.4% | 12.0% |
| 2015 | -0.4% | 1.2% |
| 2014 | 5.9% | 13.5% |
| 2013 | 1.9% | 32.3% |
| 2012 | 10.5% | 16.0% |
| 2011 | 11.9% | 1.9% |
| 2010 | 12.9% | 15.1% |
| 2009 | -3.8% | 26.4% |
| 2008 | 25.8% | -36.8% |
| 2007 | 21.7% | 5.1% |
| 2006 | 19.0% | 15.8% |
| 2005 | 5.5% | 4.8% |
| 2004 | -5.5% | 10.7% |
| 2003 | 40.1% | 28.2% |
| 2002 | 0.0% | -21.6% |
| 2001 | 9.4% | -11.8% |
| 2000 | -4.0% | -9.7% |
| 1999 | 17.9% | 20.4% |
| 1998 | 28.8% | 28.7% |
| 1997 | 6.6% | 33.5% |
| 1996 | 10.5% | 22.5% |
| 1995 | 30.8% | 38.0% |
| 1994 | 1.5% | 0.4% |
| 1993 | 17.7% | 9.7% |
| 1992 | 5.3% | 7.6% |
| 1991 | 24.1% | 30.3% |
| 1990 | 0.7% | -3.2% |
| 1989 | 17.0% | 31.5% |
| 1988 | 11.4% | 16.4% |
| 1987* | 3.0% | -20.8% |
* Partial year.
The rules as implemented
This is the exact tree Tactfolio runs, rebalanced monthly with signals and trades at the close. Open it to inspect or copy it.
- StrategyKeller VAA-G12
- WeightEqual
- IfAt least 4 of 12 conditions
- 1/3/6/12-month weighted momentum of SPY is at most 0
- 1/3/6/12-month weighted momentum of IWM is at most 0
- 1/3/6/12-month weighted momentum of QQQ is at most 0
- 1/3/6/12-month weighted momentum of VGK is at most 0
- 1/3/6/12-month weighted momentum of EWJ is at most 0
- 1/3/6/12-month weighted momentum of EEM is at most 0
- 1/3/6/12-month weighted momentum of VNQ is at most 0
- 1/3/6/12-month weighted momentum of GSG is at most 0
- 1/3/6/12-month weighted momentum of GLD is at most 0
- 1/3/6/12-month weighted momentum of TLT is at most 0
- 1/3/6/12-month weighted momentum of HYG is at most 0
- 1/3/6/12-month weighted momentum of LQD is at most 0
Then- WeightEqual
- RankTop 1 · 1/3/6/12-month weighted momentum
- TickerSHY
- TickerIEF
- TickerLQD
- RankTop 1 · 1/3/6/12-month weighted momentum
Otherwise- WeightEqual
- IfAt least 2 of 12 conditions
- 1/3/6/12-month weighted momentum of SPY is at most 0
- 1/3/6/12-month weighted momentum of IWM is at most 0
- 1/3/6/12-month weighted momentum of QQQ is at most 0
- 1/3/6/12-month weighted momentum of VGK is at most 0
- 1/3/6/12-month weighted momentum of EWJ is at most 0
- 1/3/6/12-month weighted momentum of EEM is at most 0
- 1/3/6/12-month weighted momentum of VNQ is at most 0
- 1/3/6/12-month weighted momentum of GSG is at most 0
- 1/3/6/12-month weighted momentum of GLD is at most 0
- 1/3/6/12-month weighted momentum of TLT is at most 0
- 1/3/6/12-month weighted momentum of HYG is at most 0
- 1/3/6/12-month weighted momentum of LQD is at most 0
Then- WeightSpecified
- RankTop 1 · 1/3/6/12-month weighted momentum50%
- TickerSPY
- TickerIWM
- TickerQQQ
- TickerVGK
- TickerEWJ
- TickerEEM
- TickerVNQ
- TickerGSG
- TickerGLD
- TickerTLT
- TickerHYG
- TickerLQD
- RankTop 1 · 1/3/6/12-month weighted momentum50%
- TickerSHY
- TickerIEF
- TickerLQD
- RankTop 1 · 1/3/6/12-month weighted momentum50%
Otherwise- WeightEqual
- RankTop 2 · 1/3/6/12-month weighted momentum
- TickerSPY
- TickerIWM
- TickerQQQ
- TickerVGK
- TickerEWJ
- TickerEEM
- TickerVNQ
- TickerGSG
- TickerGLD
- TickerTLT
- TickerHYG
- TickerLQD
- RankTop 2 · 1/3/6/12-month weighted momentum
- IfAt least 2 of 12 conditions
- IfAt least 4 of 12 conditions
- WeightEqual
Sources and caveats
- Wouter Keller and JW Keuning, Breadth Momentum and Vigilant Asset Allocation (2017)
- JW Keuning, Breadth Momentum and Vigilant Asset Allocation (TrendXplorer)
- Top 2 and breadth 4 with the paper's cash fraction CF = floor(b·T/B)/T. EEM stands in for VWO, which some versions use, to keep more history.
- Signals and trades use the close of the last trading day of each month, as in the source.
Common questions#
What is VAA-G12?#
It is the twelve-asset version of Wouter Keller and JW Keuning's Vigilant Asset Allocation. It holds the two strongest of twelve global funds and moves half or all of the portfolio to the best bond fund as more of them show negative momentum.
How is VAA-G12 different from VAA-G4?#
VAA-G4 holds one of four funds and goes fully defensive when any one of them weakens. VAA-G12 holds two of twelve and steps into bonds gradually, as two, then four, of the twelve weaken.
Does VAA-G12 still work?#
Since its 2017 publication it has returned about 7% a year with a deep drawdown in 2022. The live record on Tactfolio shows how it is doing now.