The iM Seasonal ETF Switching Strategy

  • This strategy exploits the anomaly that Cyclical Sectors and Small Caps perform best from November to April, and Defensive Sectors do better from May to October during most years.
  • In this analysis only one ETF is periodically selected by a simple ranking system from the cyclical and defensive groups, respectively, and held for six months.
  • Out of the 37 six-month periods, 36 periods showed gains ranging from 0.1% to 28.1%, while only one six-month period produced a loss of -9.3%.
  • For the approximately 18.5 year period from end of Oct-1999 to May-2018 the backtest showed an annualized return of 19.8% with a maximum drawdown of -30%.
  • For an “inverted” switching strategy, when cyclicals ETFs are used for the May-October period and defensive ETFs during November-April period, the annualized return was 3.2% and maximum drawdown was -60%.

 

The Case for Seasonal Switching

In this May-2013 article, What To Sell, If Selling In May, Doug Ramsey shows that since 1926 most of the excess return of Small Caps was earned during the seasonally strong months of November through April, and that one could have sat in cash each May through October and beaten “Small Cap Buy and Hold” on a total return basis.

The article also demonstrates that since the inception of the S&P 500 sector data in late 1989 it would have been beneficial to own cyclicals (Discretionary, Industrials and Materials) during the seasonally strong market months of November through April, and then swap into defensives (Staples, Health Care and Utilities) for the May to October period.

The iM Seasonal Switching Model

ETF

The cyclical ETFs used for the analysis are:

XLY

Consumer Discretionary SPDR ETF

XLI

Industrials SPDR ETF

XLB

Materials SPDR ETF

XLK

Technology SPDR ETF

VBR

Vanguard Small-Cap Value ETF

 

ETF

The defensive ETFs used are:

XLP

Consumer Staples SPDR ETF

XLV

Health Care SPDR ETF

XLU

Utilities SPDR ETF

VIG

Vanguard Dividend Appreciation ETF

IEI

iShares 3-7 Year Treasury Bond ETF

The model was back-tested on the on-line simulation platform Portfolio123, which provides historical economic and financial data as well as extended price data for ETFs before their start dates. Only one ETF is periodically selected from each group of five ETFs by a simple ranking system and held for six month. There is no market timing or stop-loss rule in the model.

A simple ranking system is used which is based on the price changes over a short period. The notion is that ETFs which have experienced a decline over a short period will bounce back, reverting and doing better than ETFs which have not declined in this way.

In the Appendix is a listing of all the 37 completed transactions. The only loss over any of the 37 six months investment period was -9.3% for the period 10/22/2007 to 4/28/2008 during the Great Recession which officially lasted from December 2007 to June 2009.

The red graph in Figure-1 shows the performance of the model with dividends and with trading costs and slippage accounted for. The blue graph represents the buy&hold performance of SPY (the ETF tracking the S&P 500 index). The model greatly out-performed SPY which only produced an annualized return of 6.0% with a maximum drawdown of -55%. Over the full period from end of Oct-1999 to May-2018 the total return of the model is 14-times that of SPY.

width="640"/(click to enlarge)

Performance by Calendar Year

The backtest shows only positive calendar year returns and produced positive returns for each calendar year, and excess returns over SPY in 13 out of 19 years.

width="640"/(click to enlarge)

What happens if the strategy is inverted?

Figure-2 shows the performance when an “inverted” seasonal switching strategy is adopted – when cyclicals ETFs are used for the May to October period and defensive ETFs during the seasonally strong market months of November through April. Again only one ETF is selected and held for six months.

The annualized return collapses to 3.2% with a maximum drawdown of -60%. Over the full period from end of Oct-1999 to May-2018 the total return of the inverted strategy model is only about 40% of SPY.

width="640"/(click to enlarge)

What happens if more than one ETF is periodically selected?

The performance diminishes if the model holds more than one ETF. The table below lists annualized returns and percentage of winning trades for the model holding one to five positions. For all cases performance is significantly better than for buy&hold of SPY which only produced an annualized return of 6.0% over the corresponding time period.

Number of Positions

Annualized Return

Percentage Winners

1

19.8%

95%

2

13.5%

84%

3

12.5%

83%

4

12.2%

82%

5

10.8%

77%

 

Conclusion

Backtesting with historic data shows that investment returns can be vastly improved by employing a seasonal ETF switching strategy. Minimum trading effort is required as one would only switch between ETFs every six months, namely on the first trading day of the last week in April and October of each year.

The performance of this model and ETF selection will be reported bi-annually.

Appendix

Transactions (Return is without dividends)

 

Symbol

Open

Close

Days Held

Return

38

XLP

04/23/2018

37

XLY

10/23/2017

04/23/2018

182

13.69%

36

XLV

04/24/2017

10/23/2017

182

12.39%

35

XLI

10/24/2016

04/24/2017

182

16.12%

34

XLU

04/25/2016

10/24/2016

182

1.50%

33

VBR

10/26/2015

04/25/2016

182

0.87%

32

IEI

04/27/2015

10/26/2015

182

0.08%

31

XLB

10/27/2014

04/27/2015

182

7.19%

30

VIG

04/28/2014

10/27/2014

182

1.63%

29

XLK

10/28/2013

04/28/2014

182

7.30%

28

VIG

04/22/2013

10/28/2013

189

10.66%

27

XLK

10/22/2012

04/22/2013

182

0.20%

26

IEI

04/23/2012

10/22/2012

182

0.37%

25

XLK

10/24/2011

04/23/2012

182

11.93%

24

IEI

04/25/2011

10/24/2011

182

4.76%

23

XLB

10/25/2010

04/25/2011

182

14.50%

22

XLV

04/26/2010

10/25/2010

182

1.14%

21

VBR

10/26/2009

04/26/2010

182

24.35%

20

XLV

04/27/2009

10/26/2009

182

18.12%

19

XLY

10/27/2008

04/27/2009

182

17.43%

18

IEI

04/28/2008

10/27/2008

182

2.62%

17

VBR

10/22/2007

04/28/2008

189

-9.31%

16

IEI

04/23/2007

10/22/2007

182

1.93%

15

XLI

10/23/2006

04/23/2007

182

7.17%

14

IEI

04/24/2006

10/23/2006

182

2.50%

13

XLI

10/24/2005

04/24/2006

182

17.62%

12

XLP

04/25/2005

10/24/2005

182

0.81%

11

XLB

10/25/2004

04/25/2005

182

6.20%

10

IEI

04/26/2004

10/25/2004

182

2.90%

9

VBR

10/27/2003

04/26/2004

182

15.77%

8

XLP

04/28/2003

10/27/2003

182

9.24%

7

XLB

10/28/2002

04/28/2003

182

5.42%

6

IEI

04/22/2002

10/28/2002

189

8.36%

5

XLY

10/22/2001

04/22/2002

182

20.75%

4

XLP

04/23/2001

10/22/2001

182

5.31%

3

XLB

10/23/2000

04/23/2001

182

24.62%

2

XLU

04/24/2000

10/23/2000

182

13.89%

1

XLK

10/25/1999

04/24/2000

182

28.09%

 

Posted in blogs, featured
24 comments on “The iM Seasonal ETF Switching Strategy
  1. TDCARLSON says:

    This is pretty cool. What sector are we in right now? I couldn’t see that from looking at this.

    Tom C

  2. TDCARLSON says:

    The ranking system you are using here appears to have a nice edge. Have you ever run this in a way that allows you switch weekly during the six month period to the leader in the respective group (Cyclical Nov-Apr, Defensive May-Oct) based on your ranking system?

    Tom C

    • geovrba says:

      This is difficult to do in P123. One could do it in excel.

      The advantage of this switching model is that one would only have to trade every six month to beat SPY. That could be attractive to many investors.

      • TDCARLSON says:

        Yes, that makes sense. Another interesting question might be: what if you used something like MAC-US timer (on the S&P) or Composite RSP-SH timer to determine whether or not to choose between cyclical and defensive regime and choose based on your ranking system.

        Tom C

      • TDCARLSON says:

        To do it in P123 you would have to have an entry rule in a sim something like

        ( ((Month 10)) AND Ticker(“XLY XLI XLB XLK VBR”) )
        OR
        ( ((Month > 4) AND (Month < 11 )) AND Ticker("XLP XLV XLU VIG IEI") )

        with an exit rule that always exits so that a new choice is evaluated each week. Been quite a while since I've done anything in P123 but I think something like that would work.

        Tom C

        • TDCARLSON says:

          sorry, lost some characters

          ( ((Month 10) AND Ticker(“XLY XLI XLB XLK VBR”) )
          OR
          ( ((Month > 4) AND (Month < 11 )) AND Ticker("XLP XLV XLU VIG IEI") )

          Tom C

  3. randyfloyd says:

    Hi, where one the site/emails will the ETF changes be reported? I wouldn’t want to miss a signal. thanks

    • randyfloyd says:

      *where ON the site

      • geovrba says:

        The model is now in XLP. We will email and post the next signal on Oct-28-2018.

        • danw says:

          Why wouldn’t the next signal be on Oct-21-2018 for trading on Oct-22-2018 since Oct-22-2018 is 26 weeks from Apr-23-2018? Also, are you trading on the open, close, or other?

          Thanks,
          Dan

          • geovrba says:

            Trading occurs on the first trading day of the last week of October, which should be Oct-29.

        • danw says:

          I’m still not understanding the trading days. For example, you have the trading day for last October as 10/23/2017. This is not the “first trading day of the last week of October”. 10/30/2017 is the first trading day of the last week of October 2017. Can you clarify?

        • danw says:

          Sorry to beat a dead horse, but 10/30/2017 is OK by this rule, but it still used 10/23/2017 above. All of the trade dates listed above actually fall between 22 and 28 so a rule such as MonthDay>=22 & MonthDay<=28 would match your data. However, you said the next trade date is the 29th which doesn't match?

  4. tony8174 says:

    How would this strategy do using fidelity select funds?

  5. c56444 says:

    What level of access will be required for the e-mail updates? Thanks.

  6. gasman says:

    Georg, Suppose one would overlay Composite Timer SPY/IEF and only enter equity positions if Composite Timer pointed to SPY. Would that improve results and drawdowns?

  7. TDCARLSON says:

    Would results improve significantly if you tweak this by overlaying your CAPE-Shiller ratio <= 13.0 threshold (that you described during your most recent article) on to this approach — that is, be in cyclicals during May-November when this condition is in place?

    Tom C

  8. numejak says:

    This strategy is essentially a cross between the “Sell in May and go away,” notion, and the Fama-French 3-Factor Model, which states that small-cap and value stocks tend to out-perform. Cyclical stocks tend to have the lower price-book ratios and lower P/E’s that you would find in a value style stock plus the higher betas needed to beat the market. Brilliant work! If you wanted to take this strategy a step further, you might bring in Bulkowski’s Ten Bagger strategy (from the book, not his website) and really amplify your gains.

    • numejak says:

      (Rules for Ten Baggers from his book)

      Small cap stocks with less than $1B market cap
      Price per share between $1 and $5
      P/B ratio < 2.5 (value stock)
      P/S ratio < 1.5
      PCF ratio < 10 but preferable < 6
      No dividend
      Has long-term debt

      These stocks, in cyclical sectors held from Nov 1 until Apr 30, except during recessionary periods, with no more than 2% – 5% of portfolio invested in any one company.

Leave a Reply

With reference to Section 202(a)(11)(D) of the Investment Advisers Act: We are Engineers and not Investment Advisers, read more ...
By the mere act of reading this page and navigating this site you acknowledge, agree to, and abide by the Terms of Use / Disclaimer
Share This