I recently received an email with a link to an article by Greg Morris in StockCharts website discussing an interesting idea about using ETF ratios in a strategy for market timing purposes (see reference at the end of the article.) This appears to be a lower risk strategy, but it has not outperformed the market on absolute returns basis in the test period, as shown below. In addition, anyone using such strategy must be prepared to execute more trades than with simpler price series momentum strategies applied to equity index ETFs.
- Four ETF ratios are constructed: SPDR S&P MidCap 400 (NYSEARCA:MDY)/iShares 3-7 Year Treasury Bond (NYSEARCA:IEI), iShares Core S&P Small-Cap (NYSEARCA:IJR)/iShares 7-10 Year Treasury Bond (NYSEARCA:IEF), Invesco QQQ (NASDAQ:QQQ)/iShares MBS (NYSEARCA:MBB), and SPDR Gold Trust (NYSEARCA:GLD)/iShares 20+ Year Treasury Bond (NYSEARCA:TLT)
- Positions resulting from each pair are allocated 25% of available equity
- Trigger is ±4% in the one-period ROC of the 2-period smoothed ratio
- If trigger > +4%, we go flat and then long the ETF in numerator of ratio
- If trigger < -4%, we go flat and then long the ETF in denominator of ratio
- Time frame is weekly. We use commission of $0.01 share in all backtests
Backtest results: 01/02/2003 - 02/06/2019
Greg Morris extended backtests results back to 1998 using the underline indexes, but here, we start in 2003 since we are using ETFs only. Below is a table that compares ETF strategy performance (weekly time frame) to long-only 50/200 moving average cross (daily time frame) in S&P 500 total return (SPY).
The equity curves of the ETF strategy and 50/200 cross are shown below.
The ETF ratio strategy outperforms the moving average cross on risk-adjusted basis in the above test period. MAR for the strategy (CAGR/Maximum drawdown) is 0.67 versus 0.44 for the moving average cross. However, the moving average cross strategy has the advantage of dealing with only one ETF and has generated just 8 trades in the test period while the ETF ratio strategy uses 8 ETFS and has generated 62 trades in the test period.
Often risk-adjusted outperformance arises due to data-mining bias (discussed in some more detail in this article) when searching for strategies and trying many different ideas with different parameters and starting days of tests until some performance criteria are fulfilled. However, choosing the best ideas while ignoring many other ideas that did not perform well is a cause of bias. In fact, the chosen ideas may be random results due to data-mining bias.
Backtest results: 01/02/2013 - 02/06/2019
Below is a table that compares the ETF strategy performance to long-only 50/200 moving average cross in SPY total return in a period staring on 01/02/2013. Therefore, we assume that someone starts using the strategy on first trading day of 2013 and naturally cannot go back in time and establish any open positions but must wait for new signals to be generated.
The strategy significantly underperforms the benchmark on absolute returns basis and also on risk-adjusted returns basis even though it has a little less than half the drawdown. MAR (CAGR/Maximum drawdown) for the strategy is 0.39 versus 0.80 for the cross strategy.
Effect of starting date
Often users of a strategy cannot realize the longer-term performance due to unfavorable starting date. In the time-domain, which is the specific path the market follows, performance depends on starting date and may widely diverge from expected longer-term performance. These issues plague some longer-term performance charts shown to investors especially when they rely on exceptional gains realized during specific periods of time (for example, CTA results). In fact, investors may face high drawdown periods in the short term even if past results appear outstanding. This is basic probability theory, but it is often ignored in strategy performance analysis.
Although the ETF ratio strategy discussed above appears to generate consistently low drawdown, alpha delivery has been hindered in recent years. The strategy cannot outperform the long-only 50/200 cross strategy in SPY especially in recent years and despite the increase in volatility. This may be the result of crowding in relative and absolute momentum strategies using ETFs or simply the effect of data-mining bias. However, for anyone looking for returns at low drawdown, this strategy may offer some potential in periods of high market uncertainty.
Reference: Core Rotation Strategy
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it. I have no business relationship with any company whose stock is mentioned in this article.