StockViz

Keller, Wouter J., Relative and Absolute Momentum in Times of Rising/Low Yields: Bold Asset Allocation (BAA) (July 18, 2022, SSRN) builds on PAA, VAA and DAA following the same philosophy. The differences between them and some of the weaknesses of BAA is summarized here.

The author swapped out SPY (the S&P 500 ETF) with QQQ (the NASDAQ 100 ETF) to juice returns. We added the original back into our backtest for completeness. Also, we ran a parallel backtest with SPHB (High-Beta ETF) instead of EEM in the canary assets like before.

The results are just as disappointing as they were before.

As much as you might hate 60/40, at least there is no model risk.

Not to belabor the point but, practically speaking, investors are better off managing drawdowns through asset allocation (hence giving up the upside during booms) or through using simple trend-following to avoid steep drawdowns (hence incurring higher transaction costs due to whipsaws) than trying to construct a Rube Goldberg machine.

Code and charts on github.

Keller, Wouter J. and Keuning, Jan Willem, Protective Asset Allocation (PAA): A Simple Momentum-Based Alternative for Term Deposits (April 5, 2016, SSRN) successfully protects your portfolio from returns.

This paper, superseded their VAA and DAA papers. The key differences between them can be found here. The paper used the Emerging Market ETF – EEM – to be one of the canary assets. While EM equities are indeed risky, a risk-off in them doesn’t really mean a risk-off in DM equities. So, we ran a parallel backtest with SPHB (High-Beta ETF) instead of EEM.

Here too, the out-of-sample performance leaves much to be desired.

Once again, investors are better off managing drawdowns through asset allocation (hence giving up the upside during booms) or through using simple trend-following to avoid steep drawdowns (hence incurring higher transaction costs due to whipsaws) than trying to construct a Rube Goldberg machine.

Code and charts are on github.

Keller, Wouter J. and Keuning, Jan Willem, Breadth Momentum and the Canary Universe: Defensive Asset Allocation (DAA) (July 12, 2018, SSRN) is a follow up to their Vigilant Asset Allocation paper that we discussed previously. A summary of their paper with some drawbacks of their approach can be found here.

The paper used the Emerging Market ETF – EEM – to be one of the canary assets. While EM equities are indeed risky, a risk-off in them doesn’t really mean a risk-off in DM equities. So, we ran a parallel backtest with SPHB (High-Beta ETF) instead of EEM.

As with VAA, the out-of-sample performance of the strategy is abysmal. Using SPHB makes it marginally better but nothing great to write home about.

To reiterate, practically, investors are better off managing drawdowns through asset allocation (hence giving up the upside during booms) or through using simple trend-following to avoid steep drawdowns (hence incurring higher transaction costs due to whipsaws) than trying to construct a Rube Goldberg machine.

Code and charts are on github.

Investors hate drawdowns. However, pretty much every attempt at shallower drawdowns comes with costs. Transaction costs and taxes aside, the biggest one is the opportunity cost of missing out on subsequent rallies after having avoided a deep drawdown.

While drawdowns are a very visible risk, the model risk of a strategy that tries to avoid them should not be discounted. The problems with drawdowns is that they are a rare beast compared to the generally upward trending nature of markets. This makes arriving at robust parameters pretty much impossible. Fragile models do not give a lot of confidence when they underperform market beta leading investors to abandon them just before a drawdown that they are designed to avoid comes calling.

However, these problems do not prevent academics from soldering on. Keller, Wouter J. and Keuning, Jan Willem, Breadth Momentum and Vigilant Asset Allocation (VAA): Winning More by Losing Less (July 14,
2017, SSRN) is one such attempt. A summary of the paper can be found here.

We reproduced it with listed ETFs and the results are sobering.

An in-sample hero but an out-of-sample dud.

No amount of behavioral reinforcements would have been enough to stick with this model when equity markets rallied after 2022.

Practically, investors are better off managing drawdowns through asset allocation (hence giving up the upside during booms) or through using simple trend-following to avoid steep drawdowns (hence incurring higher transaction costs due to whipsaws) than trying to construct a Rube Goldberg machine.

Code and charts on github.