Sector/Industry ETF Rotation Strategy


1.    Business cycle and sector/industry rotation

The business cycle, which reflects the fluctuations of activity in an economy, can be a critical determinant of equity sector performance over the intermediate term. A typical business cycle features a period of economic expansion, followed by a period of slowing growth to the peak, and then a contraction, or recession to finally reach the trough. The cycle then repeats itself.

A sector/industry rotation strategy entails “rotating” in and out of sectors/industries as time progresses and the economy moves through the different phases of the business cycle. The strategy calls for increasing allocations to sectors/industries that are expected to prosper during each phase of the business cycle while under allocating to sectors or industries that are expected to underperform.

With the wide  availability of sector/industry ETFs, the sector/industry rotation strategy is becoming more easily accessible for institutional investors as well as individual investors. One of the most widely used strategy to take advantage of the sector/industry rotation phenomenon is the momentum strategy.

A typical business cycle lasts for several years and it usually takes months or even years to go from one stage of the business cycle to the next. As such, the performances of business cycle sensitive industries/sectors are sticky, i.e., they have momentum over the intermediate term.  According to the research by Moskowitz and Grinblatt (1999, Journal of Finance), industry momentum accounts for the most, if not all, part of the total momentum phenomenon.  The momentum strategies using individual stocks are significantly less profitable once we control for industry momentum. By contrast, industry momentum investment strategies, which buy stocks from past winning industries and sell stocks from past losing industries, appear highly profitable, even after controlling for size, book-to-market equity, individual stock momentum, the cross-sectional dispersion in mean returns, and potential microstructure influences.

On the other hand, research by Daniel and Moskowitz (2016) shows that big, infrequent crashes exist in momentum strategy, which will drag down the performance. Our own research also indicates that momentum can over-shoot in the short run which causes subsequent underperformance. At AlphaFocus, we conducted several years of extensive research to improve the existing momentum strategy and form our final sector/industry rotation strategy. Compared with the simple momentum strategy and the S&P 500 benchmark, the new strategy has a higher return, lower risk and drawdown, which result in much higher Sharpe and Calmar ratios for the portfolio.

2.    The ETF selection universe

Table 1 shows the underlying investment vehicle with 40 ETFs that form our selection universe. The major selection criteria are as follows:

    • at least 10 years of history so that a reliable quantitative analysis is possible,
    • low management fee and high liquidity,
    • covers a broad range of sectors/industries.

Table 1. ETF Selection Universe Summaries

Number Ticker Description Number Ticker Description
1 FDN First Trust Dow Jones Internet Index Fund 21 XLY SPDR Consumer Discret Sel Sect ETF
2 FXD First Trust Consumer Discretionary AlphaDEX Fund 22 KBE SPDR S&P Bank ETF
3 IBB iShares Nasdaq Biotechnology 23 KRE SPDR S&P Regional Banking ETF
4 IGE iShares North American Natural Resources 24 KIE SPDR S&P Insurance ETF
5 IGV iShares North American Tech-Software 25 XHB SPDR S&P Homebuilders ETF
6 IHF iShares U.S. Healthcare Providers ETF 26 XRT SPDR S&P Retail ETF
7 IHI iShares U.S. Medical Devices ETF 27 XES SPDR S&P Oil & Gas Equipment&Svcs ETF
8 ITA iShares U.S. Aerospace & Defense ETF 28 XOP SPDR S&P Oil & Gas Explor & Prodtn ETF
9 IYT iShares Transportation Average 29 OIH Market Vectors Oil Services ETF
10 SOXX iShares Trust – iShares PHLX Semiconductor ETF 30 GDX Market Vectors Gold Miners ETF
11 QQQ PowerShares QQQ Trust, Series 1 31 PJP Invesco Dynamic Pharmaceuticals ETF
12 SPY SPDR S&P 500 ETF Trust 32 VCR Vanguard Consumer Discretionary Index Fund ETF Shares
13 XLB SPDR Materials Select Sector ETF 33 VDC Vanguard Consumer Staples Index Fund ETF Shares
14 XLE SPDR Energy Select Sector ETF 34 VNQ Vanguard Real Estate Index Fund ETF Shares
15 XLF SPDR Financial Select Sector ETF 35 VOX Vanguard Communication Services Index Fund ETF Shares
16 XLI SPDR Industrial Select Sector ETF 36 IWV iShares Russell 3000 ETF
17 XLK SPDR Technology Select Sector ETF 37 IWB iShares Russell 1000 ETF
18 XLP SPDR Consumer Staples Select Sector ETF 38 IWF iShares Russell 1000 Growth ETF
19 XLU SPDR Utilities Select Sector ETF 39 IWD iShares Russell 1000 Value ETF
20 XLV SPDR Health Care Select Sector ETF 40 IWN iShares Russell 2000 Value ETF
3.    A brief description of the sector/industry rotation strategy

Our sector/industry ETF rotation strategy builds on the bedrock of sector/industry momentum with some significant improvements. We consider our sector/industry rotation strategy a “dynamic momentum strategy” in that it is not simply a 6-month or 12-month winner’s strategy. The key ingredients of our strategy are as follows:

    • Simple momentum: the 6 or 12-month momentum is the basic building block for the dynamic sector/industry rotation strategy.
    • Risk-adjustment: a simple 6 or12-month momentum strategy picks the winners according to the past 6 or 12-month returns without considering how they got there. In our strategy, we look at the dynamics of the return path and adjust the signal accordingly. Thus it is possible for us to pick an industry with lower returns in previous months instead of limiting the selection to industries with high historical returns.
    • Over-shooting adjustment: sometimes prices of sectors/industries with strong momentum may over-shoot, i.e., they run-up too quickly and deviate from the underlying fundamentals by too much. In that case, we would expect pull backs in prices. We designed a quantitative measure to deal with this situation so that we will not chase the “over-heated” sectors/industries.
    • Seasonality: sector/industry seasonality is another pronounced phenomenon in investing that is incorporated in our sector/industry rotation strategy.
    • Factor-timing: we built non-linear time series models to predict momentum factor returns. The factor timing model improved the hit ratio with respect to the S&P 500 from 55% for a simple momentum strategy to 60% for our sector/industry rotation strategy.

Using the above procedure, we rank 40 ETFs and pick the top 8, then equally weight them to form the final sector/industry rotation portfolio. The portfolio is rebalanced monthly. The portfolio turnover varies over time based on our factor return prediction. Some months it could have 100% turnover, while other could have 0% turnover. The average turnover is about 50% per month.

4.    Backtest performance

Table 2 shows the summary statistics of our sector/industry rotation strategy portfolio return, along with the benchmark S&P 500 total return, the selection universe average return, and the simple 6-month momentum strategy return. The backtest period is from July 1999 to May 2021 with a total of 263 months. It can be seen that the portfolio (rebalanced monthly) would have earned an annualized return of 18.2%, while the S&P 500 would have earned 7.0% so the portfolio would have beaten the SP 500 by 11.2% per annum. The portfolio turnover is about 50%. We would still get an excess return of about 8% even if we take into account of the transaction cost (assume the transaction cost is 50bps).

An interesting  observation is that the portfolio return has about a 20% higher standard deviation compared with S&P 500 total return (18.1% vs 15.2%). This typically means that the portfolio has a higher risk. However, the portfolio actually has a lower downside standard deviation (11.2% vs 11.8%), and a much lower maximum drawdown (-35.4% vs -50.9%) compared with S&P 500.  So overall, the portfolio has a lower risk, yet a much higher return. The three major performance measures, the Sharpe/Sortino/Calmar ratios, are all much higher compared with that of S&P 500 benchmark.

It can also be seen  that our sector/industry rotation strategy improved upon the simple 6-month momentum strategy in all the performance measures as well. The downside standard deviation and the maximum drawdown are lower, while the Sharpe/Sortino/Calmar ratios are all higher. The last row in Table 2 shows the percentage of months that each portfolio performed better than the S&P 500 benchmark. The ETF universe average portfolio outperformed the S&P 500 benchmark 53.2% of the time, while the simple 6-month momentum strategy outperformed the S&P 500 benchmark 54.8% of the time. Our sector/industry rotation portfolio raised the hit ratio of the simple 6-month momentum strategy by 4.5% to 59.3%. This 4.5% improvement seems not so impressive at first glance, but it improved the portfolio performance substantially by avoiding those momentum crash months.

Table 2. Portfolio Summary Statistics (1999:07-2021:05)

(Note: all stats are calculated using log returns)

Portfolio Stats SP500 TR Univ Avg Momentum Rotation
Number of  Months 263 263 263 263
Average Return 0.070 0.074 0.108 0.182
Standard Deviation 0.152 0.171 0.164 0.181
Min Return -0.184 -0.222 -0.145 -0.145
Max Return 0.121 0.159 0.132 0.252
Down Stdev 0.118 0.139 0.119 0.112
Sharpe Ratio 0.459 0.430 0.656 1.003
Sortino Ratio 0.592 0.529 0.909 1.627
Calmar Ratio 0.137 0.149 0.225 0.512
Max Drawdown -0.509 -0.495 -0.479 -0.354
Hit Ratio N/A 53.2% 54.8% 59.3%

Note that although the portfolio outperformed the benchmark S&P 500 by a large margin over the whole backtest period of almost 22 years, there are still years that the portfolio underperformed. They all happen to be in the last decade and are: 2012 (13.7% vs 14.8%), 2014 (9% vs 12.8%), 2017 (9.6% vs 19.7%), and 2019 (25% vs 27.4%).

Figure 1 shows the   cumulative continuously compounded returns over the whole backtest time periods. It is clear that our sector/industry rotation strategy consistently generates alpha for the backtest period.

Figure 1. Portfolio cumulative continuously compounded returns