Market Risk Indicators – New Experiments

A modified Three-state Risk Indicator

The investing strategy used in my “Adaptive Momentum Investing” marketplace service determines the state of the market based on the difference in total return of the following four ETF pairs: (DBB, UUP), (XLY, XLP), (XLI, XLU) and (SLV, GLD). Market is risk-off if at least three pairs indicate risk-off.

Source Note: The idea of ​​using the pairs (DBB, UUP), (SLV, GLD) and (XLI, XLU) as market risk-off indicators was first published in a post by Peter Guenther and others in a Quantopian Forum thread.

Please see this Seeking Alpha article for a detailed description of the current strategy.

In this article, I investigate some further refinements of the strategy. As will be seen from the simulation results, the modified strategy does not perform substantially different from the current strategy. Our conclusion, as it will be discussed further, is that the two variants are getting similar results, and there is little change in using either of the two at different times.

The premise of this investigation is that because the (DBB, UUP) pair has the best historical performance, it may be advisable to give this pair more weight in the aggregated decision. One idea which is tested here, is to make the following two exceptions when the (DBB, UUP) pair is risk on: (1) allocate 50% to equities when only (DBB, UUP) is risk on, and (2) allocate 100% to equities when (DBB, UUP) and (SLV, GLD) are risk on.

Simulation Results

The first two tables and two graphs below show the simulation results for the simplest one-fund portfolio. Specifically, during risk on periods all the equities allocation goes to SPY.

The following two tables and two graphs show the simulation results for a portfolio investing in the top two funds from the following list: SPY, QQQ, XLF, XRT, ranked by momentum.

The first table shows the performance of the two variants versus the buy and hold strategy of the single SPY fund over the 2008-2014 period. The modified strategy has an annual return of 21.43%, that is lower than the base strategy by 1.48% annually. Compared to the buy and hold, both variants performed very well.

The chart shows that the relative performance of the two strategies is not systematically biased toward one variant. The modified variant outperformed until the Fall of 2011 correction, when the modified variant went 50% equities while the base strategy went, correctly, 100% risk-off.

2008 – 14	CAGR	stdev	maxDD	Sharpe Ratio	Sortino Ratio
Current Strategy	22.91%	14.45%	-13.38%	1.57	2.24
Modified Strategy	21.43%	14.53%	-13.35%	1.47	2.10
Benchmark SPY B&H	7.46%	22.37%	-52.36%	0.32	0.40

The second table shows the performance of the two variants versus the buy and hold strategy of the single SPY fund over the 2015-2022 period. The modified strategy has an annual return of 17.63%, that is higher than the base strategy by 0.63% annually. Compared to the buy and hold, both variants performed very well.

The chart shows that the relative performance of the two strategies is not systematically biased toward one variant. The modified variant outperformed until the Fall of 2018 correction, when the modified variant went 50% equities while the base strategy went, correctly, 100% risk-off.

The modified variant has outperformed again since the March 2020 correction, so it has a small advantage over the whole period.

2015 – 22	CAGR	stdev	maxDD	Sharpe Ratio	Sortino Ratio
Current Strategy	17.00%	14.82%	-11.82%	1.44	1.79
Modified Strategy	17.63%	12.30%	-12.45%	1.43	1.77
Benchmark SPY B&H	12.76%	17.70%	-33.72%	0.71	0.82

The last two tables show the summary performance of the portfolio of these funds: SPY, QQQ, XLF, XRT. The relative performance of the two variant strategies are following exactly the same behavior as that of the single SPY portfolio.

Compared to the single SPY portfolio, the four-fund portfolio has obtained significant increase in annual returns, all due to favorable momentum selection. The maximum drawdowns of the four-fund portfolio are a little bigger than that of the single SPY fund, but not significantly.

2008 – 14	CAGR	stdev	maxDD	Sharpe Ratio	Sortino Ratio
Current Strategy	25.76%	17.43%	-16.82%	1.48	2.10
Modified Strategy	24.81%	17.61%	-16.82%	1.41	2.02
Benchmark SPY B&H	7.46%	22.37%	-52.36%	0.32	0.40

2015 – 22	CAGR	stdev	maxDD	Sharpe Ratio	Sortino Ratio
Current Strategy	22.12%	14.05%	-12.16%	1.62	2.14
Modified Strategy	23.30%	14.62%	-13.37%	1.59	2.10
Benchmark SPY B&H	12.76%	17.70%	-33.72%	0.71	0.82

Conclusions

Here are the main conclusions.

• The modified variant outperforms most of the time, but fails in a few instances when a serious correction develops, as was the case in August-October 2011 and October-December 2018.
• Overall, there are no significant performance differences between the base and the modified variants. Both achieve excellent results compared to a buy and hold strategy.