No. of Recommendations: 59
I have posted before on some ETF sector rotation strategies 
but I thought I'd put this concept out there as it seems to work
for me.


Some observations- when you look at sector rotation,
(I used the 9 posted below) when the sectors start
improving in rank, they normally keep improving
for a while. And just as important, vice-versa.

So doing some work in EXCEL and using their 
Solver addin module, I found that using the following
rules, you end up with a pretty nice little system
capturing the leaders and the ones that seem to
be holding their own and not fading (yet).

1) First, we must be in a favorable overall market.
In order determine whether to be out of market when the "tide
is falling" I use the ratio system on the SPY ETF that I posted
a few years back- see post 241906. (You can use you own bear catcher
if you prefer). Since I have no bonds to switch
to here, I just go to cash when it says to do so.

Poor market= cash or your favorite substitute.

2) Now, assuming a "good SPY market"
take the 9 sectors ETFs, and rank them (1 the best
to 9 the poorest ranking) based on their last 8 months performance.

3) From that, look at the best 6 and choose only the
ones that have improved or maintained ranking from
5 months ago.

Example:  if XLE is currently ranked #3 of the 9, I would keep
it only if its ranking 5 months ago was not #2 or #1.  That would
mean that even it is a strong ranking now, it might be on its
way out and I would not want to keep it for the next month.

Sometimes, I would pick a top performing ETF, only to find
that its rotation time had come and it was not a darling for long.

Likewise, if say XLU was currently ranked #5 and was #8 ranked
5 months ago, I would buy it. (In the top 5 and moving up- what we want)

The takeaway here is look for strong stocks that still have
some muscle.

(Oh, yes, I do this monthly and rebalance).

Some results-
If we start with $10K in this portfolio strategy,
we end up with about $85K. B&H on the SPY doubles
your investment.


 Ending	  Port       Return  	  SPY ETF  	  Return  
Mar-17	 85,728.20 	23%	 235.74 	19%
Mar-15	 69,844.24 	38%	 198.03 	37%
Mar-13	 50,563.08 	24%	 144.49 	23%
Mar-11	 40,731.95 	59%	 117.16 	73%
Mar-09	 25,587.89 	14%	 67.54 	       -41%
Mar-07	 22,480.13 	22%	 115.28 	25%
Mar-05	 18,366.90 	54%	 92.39 	        44%
Mar-03	 11,904.22 	19%	 64.09 	       -40%
Dec-99	 10,000.00 		 106.41 

I'm not sure if this is too data mined, but moving the
criteria around a bit does not bring the system's
results down. (And probably some of its strength comes
from using the ratio system to be out of a weak market.)

Another positive- looking at a 12 month rolling return
(jan 2017/ jan 2016, dec 2016/ dec 2015...dec 2000/ dec 1999) the smallest
decrease is -7%.

MATERIALS	 XLB 
ENERGY	         XLE 
FINANCIAL	 XLF 
INDUSTRIAL	 XLI 
TECHNOLOGY	 XLK 
CON. STAPLES	 XLP 
UTILITIES	 XLU 
HEALTH CARE	 XLV 
CON. DISCRET.	 XLY
Print the post Back To Top
No. of Recommendations: 1
Some observations- when you look at sector rotation,
(I used the 9 posted below) when the sectors start
improving in rank, they normally keep improving
for a while. And just as important, vice-versa.


I like this strategy and apply it to Quant Asset Allocation Rotation.

Interesting how this was received. I mentioned it in a recent post and was told to shut-up and go away.

GD_
Print the post Back To Top
No. of Recommendations: 0
Help me understand how to pick the correct ETF. I'm confused. Ranking numbers are just for illustration purposes

ETF 8 m 5m 1m Ranking
XLB 4 4 4
XLE 5 9 8
XLF 1 1 9
XLI 3 2 7
XLK 2 3 3
XLP 9 7 5
XLU 7 5 1
XLV 8 8 6
XLY 6 6 2
Print the post Back To Top
No. of Recommendations: 15
>>Help me understand how to pick the correct ETF. I'm confused.<<

Lemme take a shot at this. (PLEASE, someone correct me if I am wrong!)

First of all, ignore the 1 month ranking, it’s irrelevant. Rank the nine ETFs based on their total return over the past 8 months (168 days). Once you have that list, throw away the bottom three, we’re only interested in numbers 1-6. In your example, that would be:

1 XLF
2 XLK
3 XLI
4 XLB
5 XLE
6 XLY

Now, you need to time travel to five months ago, in this case to the last day of October 2016. Check the 168-day total return of the nine ETFs on October 31 and rank accordingly. Those rankings should then be compared to your current list, and you would buy only those ETFs with a same or higher ranking. Hold those ETFs for one month, and then repeat the whole process on the first day of each month.

Hope that helps!
Cathy
Print the post Back To Top
No. of Recommendations: 3
Yes, Cathy has described my post's methodology correctly.
Thanks!
Print the post Back To Top
No. of Recommendations: 0
Is there a website you use to get data for the last 8 months or do you just use the closing price from the end of the month and track? This looks curiously interesting...doc
Print the post Back To Top
No. of Recommendations: 0
One of the services I subscribe to has suggested that we are in the 3rd wave (Elliots) and there is more upside to this market. I saw where you are taking questions on Elliot wave theory - what do you think about the market being on the 3rd wave and still having legs up?
...doc
Print the post Back To Top
No. of Recommendations: 0
So I did some actual charting for the current time period and I am getting U, K, and Y as 3 etf's that would be on the buy list for the current period. Am I doing this correctly? FYI the US just launched a missile strike against Syria so I'm guessing it might not be a good time to jump in the market...doc
Print the post Back To Top
No. of Recommendations: 12
>>Is there a website you use to get data for the last 8 months or do you just use the closing price from the end of the month and track?<<

You can use this link:

http://stockcharts.com/freecharts/perf.php?xlb,xle,xlf,xli,x...

Just remember to change the slider thingy at the bottom from 200 to 168, and voila!

Cathy
Print the post Back To Top
No. of Recommendations: 3
I saw where you are taking questions on Elliot wave theory - what do you think about the market
being on the 3rd wave and still having legs up?


You must have seen that on another site. I just did a search and I could not find a mention of the
Elliot wave theory on this site in the last 10 years.

The book Evidence-Based Technical Analysis uses the Elliot Wave Theory as an example of a polynomial
that can fit any historical pattern but cannot be shown in backtests to provide any predictive power.

RAM
Print the post Back To Top
No. of Recommendations: 0
Thanks, I used the bar chart and get financials, tech, industrials and consumables. So if that is the case - am I doing this right?

doc
Print the post Back To Top
No. of Recommendations: 0
You must have seen that on another site. I just did a search and I could not find a mention of the
Elliot wave theory on this site in the last 10 years.


Yes, it came from a subscription I have...doc
Print the post Back To Top
No. of Recommendations: 0
tpoto
my mistake, I looked through your messages and you were not posting about Elliot wave. thanks for the tutorial on this. I'm going to follow this until the Syria thing cools down before getting in...doc
Print the post Back To Top
No. of Recommendations: 10
I'm going to follow this until the Syria thing cools down before getting in...doc

Basing investment decisions on the news of the day is a sure way to lose money.

Elan
Print the post Back To Top
No. of Recommendations: 0
Some results-
If we start with $10K in this portfolio strategy,
we end up with about $85K. B&H on the SPY doubles
your investment.


Ending Port Return SPY ETF Return
Mar-17 85,728.20 23% 235.74 19%
Mar-15 69,844.24 38% 198.03 37%
Mar-13 50,563.08 24% 144.49 23%
Mar-11 40,731.95 59% 117.16 73%
Mar-09 25,587.89 14% 67.54 -41%
Mar-07 22,480.13 22% 115.28 25%
Mar-05 18,366.90 54% 92.39 44%
Mar-03 11,904.22 19% 64.09 -40%
Dec-99 10,000.00 106.41



Why only odd year dates? B&H doubled the investment- is this correct?
Print the post Back To Top
No. of Recommendations: 0
Basing investment decisions on the news of the day is a sure way to lose money.

There you go, I'm good at losing money - like most investors. Just trying to turn it around...doc
Print the post Back To Top
No. of Recommendations: 3
You can use this link:

http://stockcharts.com/freecharts/perf.php?xlb,xle,xlf,xli,x......

Just remember to change the slider thingy at the bottom from 200 to 168, and voila!

Cathy

http://stockcharts.com/freecharts/perf.php?XLB,XLE,XLF,XLI,X...
Print the post Back To Top
No. of Recommendations: 1
and then you switch the view to histogram :)

...doc
Print the post Back To Top
No. of Recommendations: 13
jamesmw asks

 Why only odd year dates? 

My eyes glaze over when looking at too long of a list!

But actually, this tells a better story-
how the sector rotation strategy does in the 
bull and bear markets-

	From /TO    SECTOR 	SPY B&H	 Market
	End of	    Return 	Return	 TREND 
Mar-09	Mar-17	    235%	249%    Bull 
Oct-07	Mar-09	     -3%	-47%	Bear
Mar-03	Oct-07	    121%	98%	Bull
Dec-99	Mar-03	     19%	-40%	Bear
				
				
 SPY B&H doubled? 

Yes!
But this strategy, up 8X
with a mid-teens max drawdown.
Print the post Back To Top
No. of Recommendations: 0
What are the picks for April with EOM March? Using the stockchart link Cathy posted I get XLE and XLB.
Anybody care to code this in GTR1?
Print the post Back To Top
No. of Recommendations: 0
Why 5 month and 8 months back? Why not 3 months and 6 months back? Just curious if these numbers get better results when back testing. The other interesting thing is that from 3/09 to present it looks like B&H got better returns than sector rotation, however during the previous cycles it performed better - especially the bear cycles. The bear cycles are where you really make money it seems. If you have $1000 and lose 50% =$500. Then when you grow your account 100% you are just getting back to $1000. If you lose 3%, then grow your account 100% =$1940 - the big loss is devastating. Sorry, I know I'm explaining this to super smart financial people here based on the threads but thinking out loud...doc
Print the post Back To Top
No. of Recommendations: 0
Why 5 month and 8 months back? Why not 3 months and 6 months back? Just curious if these numbers get better results when back testing. The other interesting thing is that from 3/09 to present it looks like B&H got better returns than sector rotation, however during the previous cycles it performed better - especially the bear cycles. The bear cycles are where you really make money it seems. If you have $1000 and lose 50% =$500. Then when you grow your account 100% you are just getting back to $1000. If you lose 3%, then grow your account 100% =$1940 - the big loss is devastating. Sorry, I know I'm explaining this to super smart financial people here based on the threads but thinking out loud...doc

Or was 5 and 8 months picked because that was the age of the kids at the time?
Print the post Back To Top
No. of Recommendations: 0
Why 5 month and 8 months back? Why not 3 months and 6 months back?

Good question and it is never a good practice to accept any exact numbers presented without
backtesting the concept to verify the numbers chosen are what some here call the mound of
toast. For example does it still work with 4 or 6 months with 7 or 9 months. GTR1 allows you
do very parameters to verify you haven't just discovered a statistical anomaly.

RAM
Print the post Back To Top
No. of Recommendations: 0
"allows you do very parameters"
I don't know how I did that, should have been.
allows you vary parameters

RAM - messed up
Print the post Back To Top
No. of Recommendations: 0
thanks, GTR1?

doc
Print the post Back To Top
No. of Recommendations: 0
thanks flaram, I found it...doc
Print the post Back To Top
No. of Recommendations: 2
Or was 5 and 8 months picked because that was the age of the kids at the time?

It's difficult to have kids three months apart. :-)

DB2
Print the post Back To Top
No. of Recommendations: 8
Or was 5 and 8 months picked because that was the age of the kids at the time?


It's difficult to have kids three months apart. :-)



Only for one of the parents......
Print the post Back To Top
No. of Recommendations: 0
Only for one of the parents......

big LOL...doc
Print the post Back To Top
No. of Recommendations: 8
It's too bad that my original post
(that I spent a lot of time on)
has degraded to the point of silliness of
the last few posts.
I have not posted for over a year
until recently, so I guess it's
time to take some time off again.
I've tried to help those that find
GTR1 difficult and want some simpler
methods, but I guess that this is
not what this board wants.
Print the post Back To Top
No. of Recommendations: 27
It's too bad that my original post
(that I spent a lot of time on)
has degraded to the point of silliness of
the last few posts.
I have not posted for over a year
until recently, so I guess it's
time to take some time off again.
I've tried to help those that find
GTR1 difficult and want some simpler
methods, but I guess that this is
not what this board wants.

I for one enjoy your post and find them helpful.
Print the post Back To Top
No. of Recommendations: 32
please don't let personal nonsense keep you from posting. i've done extensive searches of posting histories, and you are among the most vital and interesting in the group. please post MORE not less!
Print the post Back To Top
No. of Recommendations: 1
tpoto -

there is definitely some tuning going on as i change parameters on portfolio123, BUT
using my timing mechanism to hedge (combo of unemployment and sma of the s&p) and
working with these rules:

ticker("XLB, XLE, XLF, XLI, XLK, XLP, XLU, XLV, XLY")
forder("(close(0) - close(70)) / (close(70) - close(160))",#previous) < 5

i get sine 1/2/99:

screen: 927% cagr 13.60% mdd -16.19%
s&p500: 91% cagr 3.61% mdd -56.26%
Print the post Back To Top
No. of Recommendations: 14
What I miss is people posting some mechanical investing ideas!
http://boards.fool.com/make-mi-great-again-32609603.aspx?sor...

It's too bad that my original post (that I spent a lot of time on) has degraded to the point of silliness of the last few posts.

I would also like to see more active discussions of more ideas.

About 4 years ago, tpoto suggested {VG-Zebra}. From this, 88 posts created the similar {VG-Horse}. I now successfully use a slightly modified version of {VG-Horse}. Thank you tpoto!
http://boards.fool.com/new-screen-votes-needed-30505987.aspx...
Print the post Back To Top
No. of Recommendations: 12
well count me in as one who looks forward to topto posts
I use MI only for the general market, "timing" if you want to call it that, though I consider it as a way of recognition, not prediction. Probability related.
And a couple your indicators are on my short list.
Print the post Back To Top
No. of Recommendations: 5
we love your analysis and posts
Print the post Back To Top
No. of Recommendations: 6
Yes, if you look at the recs and not the insignificant responses (which get zero recs), you'll see you are a very valued member of this community. It's a shame that great posts get lost in the noise.

And with that, I will bow out gracefully (or not) and thank you for your past generosity. I look forward to learning even more in the future from you.

Cheers,

SD
Print the post Back To Top
No. of Recommendations: 12
<< It's too bad that my original post (that I spent a lot of time on)has degraded to the point of silliness of the last few posts. I have not posted for over a year until recently, so I guess it's time to take some time off again. >>

tpoto, I enjoy your posts, but....

....this is a sentiment that I've heard repeatedly on various message boards. Posters take on a self-proclaimed victim status: "I've spent so much time and so much thought on this post, but nobody seems to value me enough!"

This is the Internet. These are free and open message boards. You are free to post anything you want here. But "we" are free to value--or not value--your messages.

Also, you can start a thread, but you don't own it. Sometimes they take on a life of their own, and meander as they will. That's half the fun of these posting here.


Alan
Print the post Back To Top
No. of Recommendations: 2
I posted this new thread:

http://boards.fool.com/beating-spy-in-the-recent-bull-market...

because of this thread, and credited the OP, though not with a link.
Print the post Back To Top
No. of Recommendations: 6
Alan-
I looked at what you have posted to this board-
nothing of note.

Maybe when you spend a few hours putting up
something for others to hopefully benefit from,
then you might change your tune.
Print the post Back To Top
No. of Recommendations: 85
<< Alan-I looked at what you have posted to this board- nothing of note.
Maybe when you spend a few hours putting up something for others to hopefully benefit from, then you might change your tune.
>>

tpoto,

Glad you took the bait..... ;-)

I'm (one of) the OB's (Original Backtesters) here. 18 years ago, GrandPoobah (remember him?) and I did the first backtest on the PEG screen using the complete VL database. I had access to all the historical data because I'm the only one with a valid VL license for their data (1986-1999, at that time).

Since I was the owner of that data, I performed all backtests here for a year or so.

There has been a group of us working mostly "behind the scenes" who have spent tens of thousands of dollars (literally) and hundreds of hours (at least) purchasing the monthly VL data and collating it with the CRSP price data.....and making the VL backtester available to everybody for free. And we've kept this up for decades.

But whether or not any of that work was "of note" is of no interest to me. We did that work to make this board a rich environment in which to practice mechanical investing.


Alan
Print the post Back To Top
No. of Recommendations: 7
Alan wrote:

"We did that work to make this board a rich environment in which to practice mechanical investing."

Have a rec. Heck, I wish I could give you a few hundred recs. I appreciate the work that you and others have done over the years to help build our backtesting capabilities.
Print the post Back To Top
No. of Recommendations: 0
Wow - it's been awhile since there was truly a WTF post out here. Congrats!
Print the post Back To Top
No. of Recommendations: 6
Since I was the owner of that data, I performed all backtests here for a year or so.

Oh, yes, I remember studying at the alevine school of simple screens. It was time well spent....

DB2
Print the post Back To Top
No. of Recommendations: 0
I think I've been using an ETF sector rotation screen that tpoto developed, but now I can't even find the original thread. (I can't believe TMF still doesn't have functional search.)

Here's the GTR link I use (can't find the source on the MI board any more):

http://gtr1.net/2013/?~ETF%20Rotation%20PO30%2F150%20RRS63%2...

This year, I cut my exposure to this screen in half, because it's underperformed RSP since I started using it several years ago.

Does the new sector rotation developed in this thread backtest better than the older one above? I couldn't find the GTR link in the thread. Can that be posted please?

Today is yet another day when my MI positions are red while RSP is black. I am REALLY getting fed up.
Print the post Back To Top
No. of Recommendations: 1
I think I've been using an ETF sector rotation screen that tpoto developed, but now I can't even find the original thread.

Your link looks very familiar and probably not tpoto's style.

If you want a possible improvement(3pts) on performance then try adding a Hold-Til-Drop


http://gtr1.net/2013/?~ETF%20Rotation%20PO30%2f150%20RRS63%2...

Create [SMARatio]: [[Simple Moving Avg of closing g-prices over 30 days; lag=1 days]/[Simple Moving Avg of closing g-prices over 150 days; lag=1 days]]
step0: [SMARatio] > 1
step1: [RRS over 84 days; lag=1 days] Top 2
step2: [RRS over 63 days; lag=1 days] Top 2; Cash When Fewer
Holding period = 5 mkt days
Hold while passing step1
Hold for at least 4 holding periods
Rebalance to equal weight among new positions only
All trades at market close.
GD_
Print the post Back To Top
No. of Recommendations: 0
I am always concerned with testing HTD using GTR1. My understanding is that it does not matter what start date is specified, the test starts from the earliest date and results are shown from the specified start date.

Therefore with hold till drop, the sequence is set at the beginning of the backtest and not at the start of the specified period. This results in only one sequence of testing hold to drop. The results from the HTD are purely dependent on the stocks held at the very first days of the test. In practicality, you may hold different stocks that what the backtest is holding.

Craig
Print the post Back To Top
No. of Recommendations: 1
Create [SMARatio]: [[Simple Moving Avg of closing g-prices over 30 days; lag=0 days]/[Simple Moving Avg of closing g-prices over 150 days; lag=0 days]]
step0: [SMARatio] > 1
step1: [RRS over 84 days; lag=0 days] Top 2
step2: [RRS over 63 days; lag=0 days] Top param0; Cash When Fewer
Holding period = 5 mkt days
Hold while passing step1
Hold for at least 4 holding periods
Rebalance to equal weight among new positions only
All trades at market close.

I am always concerned with testing HTD using GTR1. My understanding is that it does not matter what start date is specified, the test starts from the earliest date and results are shown from the specified start date.

Therefore with hold till drop, the sequence is set at the beginning of the backtest and not at the start of the specified period. This results in only one sequence of testing hold to drop. The results from the HTD are purely dependent on the stocks held at the very first days of the test. In practicality, you may hold different stocks that what the backtest is holding.


This is almost unbelievable. Can someone confirm that those steps result in bogus GTR1 Stats.

GD_
Print the post Back To Top
No. of Recommendations: 2
I am always concerned with testing HTD using GTR1. My understanding is that it does not matter what start date is specified, the test starts from the earliest date and results are shown from the specified start date.

Therefore with hold till drop, the sequence is set at the beginning of the backtest and not at the start of the specified period. This results in only one sequence of testing hold to drop. The results from the HTD are purely dependent on the stocks held at the very first days of the test. In practicality, you may hold different stocks that what the backtest is holding.


That is a constraint, but except for a fairly short backtest period (whose validity is open to question because the period is short), I really don't see it as much of a limitation. No data . . . just saying.

But if it really bothers you, I think there is a way to get exactly what you want.
I don't have time to write it up (primarily because my gtr1-code-from-scratch skills are severly challenged), but here is my thought process.

Pick a security that predates your desired start time, and create a custom universe of that security; i.e. {XXX}
Create a BCC-like flag for that security along the following lines:
Find N, where Backtest_start_date = XXX_start_date + N
Then write a backtest that will be the basis for the URL of an "imports" function
Start_flag: if(dspo(1) >= N, 1, 0)
step0: Start_flag < 0 [dummy step that passes no securities, because gtr1 requires at least one step]

Create the URL for the Start_flag signal function that can then be imported into your backtest with the "Close all liquid positions and hold cash . . ." box set to Start_flag = 0.
I could be wrong, but I think this should keep the backtest is cash until your specified start date is reached.

There may be other (better) ways to bypass the "start date" constraint, but hopefully you get the idea.

Charlie
Print the post Back To Top
No. of Recommendations: 5
I agree with WaveDoc; Except for a fairly short backtest this shouldn't matter. The daily start statistics
over the full period are the most important.

From rgearyiii correcting my poor attempt at changing start dates:

If you want to force the backtester to buy a full set of fresh stock picks on a certain date, then you can do that by forcing it to hold cash before that date with a step like

linear(1,ord(1),-1,date2ord(20131231)) >= 0

Translation: [1*[Mkt Date as Ordinal; lag=1 days] - 1*[20131231 as market date ordinal 23373]] >= 0

which allows nothing to pass the screen until 20140102 (assuming lag adjustments are zero or left to default). If the holding period is set to 252 market days, then all cycles hold cash for the entire backtest before 2014012, and don't purchase stocks until their first scheduled trading date in 2014 arrives. That means that averaged performance statistics do not reflect all cycles holding stock picks with a report starting date any earlier than the end of 2014. This is in contrast to a backtest of a regular screen that allows stocks to pass on day one of the backtest, where all cycles are initialized with the same holdings as cycle 0 until their first scheduled trading date arrives.


RAM
Print the post Back To Top
No. of Recommendations: 2
GD

I don't think you get bogus GTR1 stats, I just think that you may not be getting what you think you are getting.

example

Earliest start date 20010105

Hold top 2 until drop below 5. 5 day hold

Stock ranks

A, B, C, D, E. Selects A and B

20010112

Ranks F, C, A, B, D. Continues to hold A and B under HTD rule.

Now, if you entered the start date as 20010112, you may think you would be picking F and C as you are not currently owning any stocks. I think GTR1 is still selecting A and B as GTR1 starts at the earliest possible start date. So your HTD screen depends on the earliest start date and you can't determine how effective HTD is with other start dates, unless you do something fancy to close all positions. There may be a significant difference starting with A and B compared to starting with F and C, even though the start dates are 1 week apart.

Craig
Print the post Back To Top
No. of Recommendations: 0
I don't think you get bogus GTR1 stats, I just think that you may not be getting what you think you are getting.

thanks for the reply

Its now clear as mud.

GD_
Print the post Back To Top
No. of Recommendations: 1
FLARAM:
From rgearyiii correcting my poor attempt at changing start dates:

If you want to force the backtester to buy a full set of fresh stock picks on a certain date, then you can do that by forcing it to hold cash before that date with a step like

linear(1,ord(1),-1,date2ord(20131231)) >= 0

Translation: [1*[Mkt Date as Ordinal; lag=1 days] - 1*[20131231 as market date ordinal 23373]] >= 0


RAM - I had forgotten about the date2ord function - much cleaner.
Actually, the date2ord discussion has been languishing at the bottom of the Dummies manual with a note to self to find a place for that discussion. Thanks for the reminder; it now has a home.

Charlie
Print the post Back To Top
No. of Recommendations: 2
BAGoldmn - Today is yet another day when my MI positions are red while RSP is black. I am REALLY getting fed up.

The US Broadmarket has been in a "Trading Range" for about 6 weeks so don't expect too much from Asset Rotation screens until some things change.
If you are experiencing whiplash with the RRS63 top 2 then reconsider going with the additional smoothing suggestion (RRS63 HTD RRS84).

http://gtr1.net/2013/?~ETF%20Rotation%20PO30%2f150%20RRS63%2...

GD_
Print the post Back To Top
No. of Recommendations: 4
MapG why use a complex (RRS63 HTD RRS84) when a simpler RRS84 top2 HTD top 3 does a better job
of containing whiplash plus it has a higher CAGR, lower turnover, lower UI and lower MDD?

              Top2htd3  Top263/84
CAGR: 16.2 14.9
TR: 444.4 378.3
GSD(20): 10.9 12.0
GDD(20; 0%): 6.8 7.6
GDDD3: 4.1 5.7
MDD: -14.2 -19.2
UI(20): 3.9 5.6
Sharpe(20): 1.4 1.2
Beta(20): 0.2 0.2
TI(20): 74.3 62.5
AT: 1.8 3.1


http://gtr1.net/2013/?~ETF%20Rotation%20PO30%2f150%20RRS63%2...
Print the post Back To Top
No. of Recommendations: 2
MapG why use a complex (RRS63 HTD RRS84) when a simpler RRS84 top2 HTD top 3 does a better job
of containing whiplash plus it has a higher CAGR, lower turnover, lower UI and lower MDD?


I like it.

GD_
Print the post Back To Top
No. of Recommendations: 1
top 2 rrs84 HTD top 3 rrs84

Symbol Company Name
VUG VANGUARD GROWTH ETF
VGK VANGUARD FTSE EUROPE
VNQ VANGUARD REIT ETF
http://gtr1.net/2013/?~ETF%20Rotation%20PO30%2f150%20RRS63%2...

Pretty much consistant with my asset rotation screen holdings

GD_
Print the post Back To Top
No. of Recommendations: 2
Pretty much consistant with my asset rotation screen holdings

Yes, Exactly the same I was just trying to point out the advantages and simplicity of hold 2 till drop 3.

I have been backtesting many of the popular rotation schemes that use absolute and/or relative momentum,
volatility and/or correlation elements. All that I have been playing with that were developed more than 3 years
ago haven't done well post discovery. On the other hand none of them has actually crashed it's just that they
haven't matched a simple equal weight. allocation. Probably foolish of me to get upset with only a few relatively
poor years.

RAM
Print the post Back To Top
No. of Recommendations: 2
You guys have lost me. The last GTR link is identical to the one I posted as far as I can tell. I have no idea what your various hold-till-drop acronyms mean. I trade once a month.
Print the post Back To Top
No. of Recommendations: 1
You guys have lost me. The last GTR link is identical to the one I posted as far as I can tell. I have no idea what your various hold-till-drop acronyms mean. I trade once a month.

The logical steps for that may help your understanding of that URL

Criteria:
HTDIF: ( Held < 31d ),
SELLIF: ( RRS84_Rank > 3 or SMA30 < SMA150),
BUYIF: ( RRS84_Rank <= 2 ),
HTDIF no change

Create [SMARatio]: [[Simple Moving Avg of closing g-prices over 30 days; lag=1 days]/[Simple Moving Avg of closing g-prices over 150 days; lag=1 days]]
step0: [SMARatio] > 1
step1: [RRS over 84 days; lag=1 days] Top 3
step2: [RRS over 84 days; lag=1 days] Top 2; Cash When Fewer
Holding period = 5 mkt days
Hold while passing step1
Hold for at least 4 holding periods
Rebalance to equal weight among new positions only
All trades at market close.

GD_
Print the post Back To Top
No. of Recommendations: 3
You guys have lost me. The last GTR link is identical to the one I posted as far as I can tell. I have no
idea what your various hold-till-drop acronyms mean. I trade once a month.


Your "I trade once a month" based on the GTR1 link you provided actually means you check the ranking
once a month and if the ranking has changed trade. Doing it this way you actually have an annual
turnover of 3.4 and as you are holding 2 means 6.8 trades per year. This gave you a 12.9 CAGR, -20.2
MDD and an Ulcer Index of 5.9.

By contrast the method MapG and I propose requires you to check the ranking weekly but reacts faster
to changes gives you another 3.3% per year CAGR, requires 3 less trades per year, had 6% less
drawdown and an Ulcer Index much better. The question becomes for you is checking the status weekly
worth the better performance.

RAM
Print the post Back To Top
No. of Recommendations: 3
tpoto,

Glad you took the bait..... ;-)

I'm (one of) the OB's (Original Backtesters) here. 18 years ago, GrandPoobah (remember him?) and I did the first backtest on the PEG screen using the complete VL database. I had access to all the historical data because I'm the only one with a valid VL license for their data (1986-1999, at that time).

Since I was the owner of that data, I performed all backtests here for a year or so.

There has been a group of us working mostly "behind the scenes" who have spent tens of thousands of dollars (literally) and hundreds of hours (at least) purchasing the monthly VL data and collating it with the CRSP price data.....and making the VL backtester available to everybody for free. And we've kept this up for decades.

But whether or not any of that work was "of note" is of no interest to me. We did that work to make this board a rich environment in which to practice mechanical investing.


Alan


Alan, you are awesome. Thanks for all you do - behind the scenes - without attitude...doc
Print the post Back To Top
No. of Recommendations: 0
Ok I got. XL*
5/01/2017--1F,2K,3I,4Y,5B,6U,7V,8P,9E
12/1/16----1F,2E,3I,4B,5K,6Y,7V,8U,9P
What are the picks for this month?
Print the post Back To Top
No. of Recommendations: 0
Has something changed at gtr1? I have been running a paper portfolio for about a month checking the picks weekly using the link in message 266591. Now when I run it I don't get any results.

Ed
Print the post Back To Top
No. of Recommendations: 2
Has something changed at gtr1?

Yes and no.(:-
The custom universe depends on Yahoo adjusted price which thanks to Yahoo is no longer available.

GD_
Print the post Back To Top
No. of Recommendations: 0
thanks
Print the post Back To Top
No. of Recommendations: 0
Hi tpoto and others,
How does this sector strategy work if it is traded less frequently. Although some of my savings are in a account that only charges $5 a trade, I have another account which is with a full service account where trade costs are very expensive. My spouse used to work for this full service broker so this money won't be going to a lower cost broker.
Thanks,
Elliot
Print the post Back To Top
No. of Recommendations: 25
Someone pointed out this post to me so I took a few minutes to validate the results and see if the optimal values could be improved. This is based on 2/1/2000 to 11/1/2017. Based on the various lookbacks and the limitation of the data set, this is the farthest back that I could go. Also, I'm using Yahoo Adjusted monthly data.

Rank Para         4         5         6            7          8        9                           
CAGR 11.88% 13.28% 13.23% 12.99% 13.29% 13.37%
Lookback Para 4 5 6 7 8 9
CAGR 11.98% 11.83% 11.65% 11.49% 13.37% 13.97%

2/1/2000
11/1/2017
Sector No Uses No Uses No Uses No Uses
Rotation Timing Timing Timing Timing Timing Timing Timing Timing
Year Sector Rotation
$10,000.00 $10,000.00 SPY SPY $10,000.00 $10,000.00
2000 29.10% 29.10% $12,909.87 $12,909.87 0.81% 0.81% $10,081.27 $10,081.27
2001 -12.77% -1.20% $11,260.65 $12,755.55 -16.34% -7.31% $8,433.73 $9,344.12
2002 -21.35% 5.58% $8,855.96 $13,467.07 -22.76% 4.83% $6,514.52 $9,795.04
2003 33.90% 35.67% $11,857.91 $18,271.20 34.01% 36.10% $8,730.22 $13,330.97
2004 10.36% 10.36% $13,086.06 $20,163.59 6.12% 6.12% $9,264.34 $14,146.57
2005 17.07% 17.07% $15,319.30 $23,604.67 9.81% 9.81% $10,172.95 $15,534.01
2006 10.41% 10.41% $16,914.52 $26,062.64 14.83% 14.83% $11,681.61 $17,837.71
2007 9.04% 9.04% $18,443.16 $28,418.05 -2.67% -2.67% $11,369.14 $17,360.58
2008 -35.75% -13.18% $11,849.66 $24,672.23 -38.25% -15.68% $7,020.26 $14,638.53
2009 38.71% 21.41% $16,436.11 $29,954.00 32.65% 17.90% $9,312.55 $17,259.21
2010 25.01% 25.01% $20,547.06 $37,446.00 22.18% 22.18% $11,377.80 $21,086.81
2011 10.31% 10.31% $22,665.01 $41,305.87 4.19% 4.19% $11,854.83 $21,970.90
2012 10.27% 10.27% $24,993.79 $45,549.96 16.52% 16.52% $13,813.74 $25,601.40
2013 24.40% 24.40% $31,092.47 $56,664.50 21.43% 21.43% $16,773.78 $31,087.33
2014 24.11% 24.11% $38,588.72 $70,326.04 14.12% 14.12% $19,143.02 $35,478.31
2015 -10.14% -10.14% $34,677.62 $63,198.25 -0.87% -0.87% $18,976.78 $35,170.21
2016 39.72% 39.72% $48,451.82 $88,301.06 19.98% 19.98% $22,767.44 $42,195.54
2017 15.33% 15.33% $55,879.77 $101,838.13 17.99% 17.99% $26,863.20 $49,786.33
ROI 458.80% 918.38% 168.63% 397.86%
CAGR 10.18% 13.97% 5.73% 9.46%


The optimal values are illustrated at the top. It is best to consider all of the rankings, not just the best six of them. And the month lookback is best set to nine months rather than eight months.

Excellent work Tpoto. Creative as always!
Print the post Back To Top
No. of Recommendations: 5
zeelotes - i've done a lot of reading of archives here and wanted to thank you for all your work and contributions. nice to see you posting again.
Print the post Back To Top
No. of Recommendations: 2
lintness wrote:
i've done a lot of reading of archives here and wanted to thank you for all your work and contributions.

You're welcome.

Also, I've added the spreadsheet to my blog for those who would like to use the information for their own evaluation of the method. Just hit Downloads at the top, and then go to the Research section, and it is on the second page of the available downloads.
Print the post Back To Top
No. of Recommendations: 1
The optimal values are illustrated at the top. It is best to consider all of the rankings, not just the best six of them. And the month lookback is best set to nine months rather than eight months.


Looks to me that all 9 sectors meet the criteria now, so might as well buy SPY.
Print the post Back To Top
No. of Recommendations: 0
The optimal values are illustrated at the top. It is best to consider all of the rankings, not just the best six of them. And the month lookback is best set to nine months rather than eight months.


Looks to me that all 9 sectors meet the criteria now, so might as well buy SPY.
------------------------------------------------------------------
Looks like I was wrong 6 now above 9 months. Please check me on this.
Print the post Back To Top
No. of Recommendations: 2
There is only one case in the full history where eight are held - 7/1/2006. No cases with nine held. Quite a few with seven held.

Here is the full history with those months holding seven or more in bold. Right now the hold is five.

  Date     Count
2/1/2000 6
3/1/2000 4
4/1/2000 6
5/1/2000 7
6/1/2000 5
7/1/2000 6
8/1/2000 5
9/1/2000 5
10/1/2000 5
11/1/2000 6
12/1/2000 5
1/1/2001 5
2/1/2001 5
3/1/2001 4
4/1/2001 5
5/1/2001 5
6/1/2001 4
7/1/2001 6
8/1/2001 6
9/1/2001 5
10/1/2001 7
11/1/2001 6
12/1/2001 6
1/1/2002 6
2/1/2002 6
3/1/2002 5
4/1/2002 4
5/1/2002 6
6/1/2002 5
7/1/2002 6
8/1/2002 7
9/1/2002 6
10/1/2002 7
11/1/2002 4
12/1/2002 7
1/1/2003 7
2/1/2003 4
3/1/2003 6
4/1/2003 3
5/1/2003 4
6/1/2003 4
7/1/2003 5
8/1/2003 4
9/1/2003 7
10/1/2003 5
11/1/2003 4
12/1/2003 6
1/1/2004 6
2/1/2004 5
3/1/2004 6
4/1/2004 4
5/1/2004 5
6/1/2004 5
7/1/2004 5
8/1/2004 5
9/1/2004 6
10/1/2004 6
11/1/2004 4
12/1/2004 6
1/1/2005 5
2/1/2005 5
3/1/2005 7
4/1/2005 5
5/1/2005 6
6/1/2005 4
7/1/2005 5
8/1/2005 6
9/1/2005 5
10/1/2005 5
11/1/2005 5
12/1/2005 4
1/1/2006 6
2/1/2006 5
3/1/2006 7
4/1/2006 6
5/1/2006 4
6/1/2006 5
7/1/2006 8
8/1/2006 4
9/1/2006 5
10/1/2006 6
11/1/2006 5
12/1/2006 6
1/1/2007 6
2/1/2007 5
3/1/2007 5
4/1/2007 6
5/1/2007 4
6/1/2007 4
7/1/2007 5
8/1/2007 4
9/1/2007 3
10/1/2007 5
11/1/2007 5
12/1/2007 7
1/1/2008 7
2/1/2008 7
3/1/2008 7
4/1/2008 7
5/1/2008 7
6/1/2008 6
7/1/2008 6
8/1/2008 7
9/1/2008 4
10/1/2008 6
11/1/2008 7
12/1/2008 6
1/1/2009 6
2/1/2009 6
3/1/2009 7
4/1/2009 5
5/1/2009 7
6/1/2009 5
7/1/2009 4
8/1/2009 6
9/1/2009 5
10/1/2009 6
11/1/2009 5
12/1/2009 4
1/1/2010 6
2/1/2010 5
3/1/2010 5
4/1/2010 6
5/1/2010 7
6/1/2010 5
7/1/2010 5
8/1/2010 3
9/1/2010 4
10/1/2010 5
11/1/2010 5
12/1/2010 3
1/1/2011 5
2/1/2011 5
3/1/2011 4
4/1/2011 4
5/1/2011 6
6/1/2011 6
7/1/2011 5
8/1/2011 3
9/1/2011 5
10/1/2011 4
11/1/2011 5
12/1/2011 6
1/1/2012 7
2/1/2012 5
3/1/2012 4
4/1/2012 5
5/1/2012 5
6/1/2012 6
7/1/2012 4
8/1/2012 5
9/1/2012 5
10/1/2012 5
11/1/2012 3
12/1/2012 6
1/1/2013 7
2/1/2013 6
3/1/2013 6
4/1/2013 5
5/1/2013 4
6/1/2013 6
7/1/2013 6
8/1/2013 6
9/1/2013 5
10/1/2013 5
11/1/2013 5
12/1/2013 6
1/1/2014 6
2/1/2014 5
3/1/2014 5
4/1/2014 6
5/1/2014 7
6/1/2014 5
7/1/2014 6
8/1/2014 5
9/1/2014 6
10/1/2014 6
11/1/2014 5
12/1/2014 6
1/1/2015 5
2/1/2015 6
3/1/2015 6
4/1/2015 6
5/1/2015 7
6/1/2015 7
7/1/2015 5
8/1/2015 6
9/1/2015 7
10/1/2015 7
11/1/2015 6
12/1/2015 6
1/1/2016 6
2/1/2016 5
3/1/2016 5
4/1/2016 5
5/1/2016 4
6/1/2016 4
7/1/2016 6
8/1/2016 7
9/1/2016 3
10/1/2016 4
11/1/2016 6
12/1/2016 5
1/1/2017 6
2/1/2017 6
3/1/2017 6
4/1/2017 6
5/1/2017 6
6/1/2017 7
7/1/2017 6
8/1/2017 5
9/1/2017 4
10/1/2017 5
11/1/2017 5
Print the post Back To Top
No. of Recommendations: 0
how easy or difficult would it be for someone to run a hedged version where you're short the etfs that don't meet the investing criteria (dollar for dollar?)?
Print the post Back To Top
No. of Recommendations: 1
how easy or difficult would it be for someone to run a hedged version where you're short the etfs that don't meet the investing criteria (dollar for dollar?)?

Wall Street is littered with the corpses of people & funds that played on the short side. When a short goes bad, it can go really really bad.
When a long goes bad, that bad position becomes a smaller and smaller piece of your portfolio, the loss is self-limiting.
But when a short goes bad, that bad position becomes a larger and larger piece of your portfolio.

Read Jim Cramer's first (?) book where he describes trying to exit some large positions in his hedge fund.
Print the post Back To Top
No. of Recommendations: 0
Hi all, I'm new to TMF and am still wading my way through all of the knowledge entrenched here - to everyone that's contributed to it, thank you.

I'm having trouble interpreting the columns in Zeelotes's table. What variable changes between the "No Timing" and "Uses Timing" for both Sector and SPY?
Also, for this data set, I am assuming both the ranking and lookback parameters are set to "9"?

I realize this is an old thread so apologies if this has already been discussed in a different thread or on the Wiki. If so, any help finding it would be appreciated.

-BFrecks
Print the post Back To Top