Of all the ideas on this board, that seemed to me to be one of the best. Has anyone done it for years?
I havenāt used sector rotation for many years but here is an update to my Fidelity Sector Rotation
Post from 12/2/2020; TMF: Sector Rotation another check / Mechanical Investing (fool.com)
Sector Rotation for the last 10 years with a bull market not draw downs has neither helped nor hurt.
Likewise from 10/15/2007 which included the ā08 bear.
Using rgearyiiiās rrs type momentum wave and Fidelity Sectors top 3 using the following sector funds:
FNARX 1997-02-28 Fidelity Select Natural Resources Portfolio
FSPTX 1981-07-14 Fidelity Select Technology Portfolio
FSRFX 1986-09-29 Fidelity Select Transportation Portfolio
FSAGX 1985-12-16 Fidelity Select Gold Portfolio
FIUIX 1987-11-27 Fidelity Telecom and Utilities Fund
FSAVX 1986-06-30 Fidelity Select Automotive Portfolio
FSELX 1985-07-29 Fidelity Select Semiconductors Portfolio
FSPCX 1985-12-16 Fidelity Select Insurance Portfolio
FSCHX 1985-07-29 Fidelity Select Chemicals Portfolio
FSDAX 1984-05-08 Fidelity Select Defense & Aerospace Portfolio
FSHOX 1986-09-29 Fidelity Select Construction & Housing Portfolio
FSHCX 1986-06-30 Fidelity Select Health Care Services Portfolio
FCYIX 1997-02-28 Fidelity Select Industrials Portfolio
http://gtr1.net/2013/?~Fid_SectorRot:s20071015f0.15000::dspoā¦
From 10/15/2007 till present:
SectorRot SPY
CAGR: 9.3 9.9
TR: 258 284
Log2TR: 1.8 1.9
SAWR(20; 0.95): 5.8 6.7
GSD(20): 23.9 19.2
DIGSD(20; 0%): 27.9 23.4
LDD(20; 0%): 16.0 13.4
LDDD3: 16.1 12.5
MDD: -58.9 -54.8
UI(20): 17.0 14.1
Sharpe(20): 0.5 0.6
Beta(20): 1.1 1.0
TI(20): 10.0 10.4
AT: 4.3 0.0
A bust, a lot of effort for no gain. Sector rotation seemed like a good idea but not everything works
the way we expect it to.
RAM
A bust, a lot of effort for no gain. Sector rotation seemed like a good idea but not everything works the way we expect it to.
To be fair, thatās only a test of a specific way of using it, and (if Iām not mistaken) a way made up long after the idea was first discussed ad nauseam.
The versions I tested long ago were
-
Pure calendar months. It makes a difference.
-
10 month relative strength, though 9 wasnāt bad. Quite different from the fancy formula of RRS returns you have in that test.
-
More funds: top 6-8, even 10.
-
Include all the sector funds that were available at the time.
-
Optional: Add 10 āfakeā funds into the rotation that are the returns on cash.
When you sort returns, these will sort to the bottom in bull markets and sort to the top in long bear markets.
This in effect eliminates holding any fund that has lost money in the lookback period: you hold cash for that portion.
Adds a bit of value in broadly based bear markets that are long and gradual, though it hurts a bit at other times.
Opinions differ on whether thatās worthwhile. It depends on what you want to happen in a drawn out bear.
I havenāt tested this lately.
Though it did work nicely for quite a few years after the first flurry of interest.
Jim
'Not quite what Jim was suggesting, but below is a calendar month relative strength momentum strategy, with cash, using the funds in RAMcNās post:
https://www.portfoliovisualizer.com/test-market-timing-modelā¦
The stats and equity curve look good.
But this illustrates a favorite point of mine: stats and equity curves can be highly deceptive.
You need to look at consistency of performance i.e. be sure to click on ārolling returnsā.
Depending on your lookback period, the outperformance ends around 2010 to 2012.
I didnāt find any lookback period or ānumber of funds heldā that made this look good
FWIW, this is the list of funds my test used.
Original studies from 2004 by someone else (Peter D), most of my follow up analysis done in 2010-2014.
FSELX FSPTX FDCPX FSESX FSCSX FSAGX FBIOX FSTCX
FSLBX FIDSX FSPHX FSENX FSDAX FDLSX FSCHX FDFAX
FSAIX FSVLX FSPCX FSRPX FSAVX FSRBX FBMPX FSPFX
FSHOX FSCGX FSDPX FSRFX FSDCX FSHCX FSLEX FSCPX
FSNGX FCYIX FNARX FBSOX FSMEX FNINX FWRLX FPHAX
FSUTX
Main observation: it used a whole lot more funds to pick from.
The GTR1 test in this thread test used 13, the old test used 41.
Jim
Five or so of the tickers werenāt recognized. I didnāt look into it but presumably they donāt presently exist.
Using the rest, itās the same story as before: the equity curve and stats can look interesting, but can be deceptive.
Rolling returns shows no consistent outperformance:
https://www.portfoliovisualizer.com/test-market-timing-modelā¦
I used my own version for this for 6-7 years and in the end it about broke even, although for the first few years it did quite well. Would have certainly benefited from some very general timing, in or out depending on how the overall market was doing.
tedtherdog pointed out a flaw in most sector rotation schemes by adding cashx to the mix, most sector rotation systems donāt have any safe alternatives. Using GTR1 I added an version with 4 cash/bond alternatives and then an alternate momentum.
Fourteen years from 10/15/2007 which included the ā08 bear. All trades with 0.15% friction.
Using rgearyiiiās rrs type momentum wave and Fidelity Sectors top 3 using the following sector funds:
FNARX 1997-02-28 Fidelity Select Natural Resources Portfolio
FSPTX 1981-07-14 Fidelity Select Technology Portfolio
FSRFX 1986-09-29 Fidelity Select Transportation Portfolio
FSAGX 1985-12-16 Fidelity Select Gold Portfolio
FIUIX 1987-11-27 Fidelity Telecom and Utilities Fund
FSAVX 1986-06-30 Fidelity Select Automotive Portfolio
FSELX 1985-07-29 Fidelity Select Semiconductors Portfolio
FSPCX 1985-12-16 Fidelity Select Insurance Portfolio
FSCHX 1985-07-29 Fidelity Select Chemicals Portfolio
FSDAX 1984-05-08 Fidelity Select Defense & Aerospace Portfolio
FSHOX 1986-09-29 Fidelity Select Construction & Housing Portfolio
FSHCX 1986-06-30 Fidelity Select Health Care Services Portfolio
FCYIX 1997-02-28 Fidelity Select Industrials Portfolio
http://gtr1.net/2013/?~Fid_SectorRot:s20071015f0.15000::dspoā¦ā¦
With the addition of safe alternatives.
TLT iShares 20+ year treasury
SHY iShares 1-3 year treasury
IEF iShares 7-10 year treasury
CASHX Nasdaq delayed price currency in USD
http://gtr1.net/2013/?~Fid_SectorRot:s20071015f0.15000::dspoā¦
With the addition of safe alternatives but using 210 day ~ 10month total return instead of rrs.
http://gtr1.net/2013/?~Fid_SectorRot:s20071015f0.15000::dspoā¦
The results from 10/15/2007 till present:
FidSecRot SPY Fid+Safe Fid+Safe10m
CAGR: 9.3 9.9 9.1 9.5
TR: 258 284 252 269
Log2TR: 1.8 1.9 1.8 1.9
SAWR(20; 0.95): 5.8 6.7 7 7.5
GSD(20): 23.9 19.2 19.1 17.5
DIGSD(20; 0%): 27.9 23.4 21.8 19.7
LDD(20; 0%): 16 13.4 12.6 11.3
LDDD3: 16.1 12.5 10.5 9.2
MDD: -58.9 -54.8 -36.6 -29.7
UI(20): 17 14.1 10.3 9.0
Sharpe(20): 0.5 0.6 0.6 0.6
Beta(20): 1.1 1 0.7 0.5
TI(20): 10 10.4 14 18.3
AT: 4.3 0 4.8 3.3
The last option comes out ahead in SAWR, MDD, UI with only a small penalty in CAGR to SPY alone.
RAM
Actually, Jim mentioned including a cash alternative.
FWIW, this is a version of ādual momentumā, advocated by Antonacci, that some of use.
oops, meant āsome of us useā
It seems like another anomaly-thatās-changed strategy. Maybe now the rotations are too fast for us retails to leverage. There is huge money moving in and out of these things weekly if not daily.
This article declared it dead in 9/19.
https://blog.thinknewfound.com/2019/09/sector-momentum/
Anecdotally and not for nothing I did this with a portion for 3.5 years '16 to '19 and made a total return of 23%.
Adding a timing component seems to considerably reduce the maximum drawdown:
The smoothness of the equity curve is also remarkableā¦
Stefano
Adding a timing component seems to considerably reduce the maximum drawdown:
The smoothness of the equity curve is also remarkableā¦
Stefano
Run it from 2010, does not look good at all.
Love that siteā¦ and yet, shortening the backtest to 11 years shows āthings have changedā here as well - slightly worse than the S&P, without the work -