duke56468
683 posts
msg #97100
- Ignore duke56468
modified |
10/19/2010 11:17:48 PM
Kevin when I run relative strength (at ETF Replay) on the portfolio
IWM,EEM,AGG against SPY since 2003 The best performance is 397% updating quarterly and the worst performance is 130% updating weekly all compared to spy of 53.7%. Do you think this longer wait on the updates would hold true for your TSI filter also? It is hard to sit and wait but back-test says it is best.
|
Kevin_in_GA
4,599 posts
msg #97106
- Ignore Kevin_in_GA |
10/20/2010 8:26:03 AM
No disagreement on this, but remember that etfreplay.com's calculation is based on a rank multiplied by your percentage allocation, not on the actual indicator value itself. Also based on RS rather than TSI.
I've looked at both, and both are profitable. I posted my data results for VWO, IWM, SPY, and AGG earlier but will post them here again for comparison.
Since 1/3/2007, the weekly TSI(5,5,9) method - buy the top scoring ETF, sell when another one beats it checking only weekly on Fridays - gives a 59.4% return through 8/27/2010 on 70 trades. It is now up about 8% from that point, so let's call it a 65% return to take into account yesterday's down day.
Using a weekly rebalancing method (I know you can use others as well, but this is what I am looking for in terms of timing and it allows for an apples-to-apples comparison) gives the following:
1 day RS: 102%
2 day RS: -2.9%
5 day RS: 55.3%
10 day RS: 25.1%
20 day RS: 42.5%
3 month RS: 57.5%
6 month RS: 94.1%
Interesting spread for the data. Some look great, but what concerned me here is the radical difference between a 1 day RS and a 2 day RS, then again with the 5 day RS. That type of variability makes me concerned that the system as it is used here is not all that robust.
As you know, I like the concept and originally posted it here. Since then, I have looked more at TSI and am happy with it for now.
|
duke56468
683 posts
msg #97108
- Ignore duke56468 |
10/20/2010 9:28:23 AM
Thanks again for sharing all your hard work.
|
jnafach
74 posts
msg #97123
- Ignore jnafach |
10/20/2010 10:14:26 PM
Hey Kevin, I looked back on ETf and run it from early 2007 it did give better than the weekly approach you are looking at here, Am I missing anything. I agree with you that doing weekly feels better but that may not really works better.
It is similar to approach I was looking looking at allocating 20% between VTI, VEU, IEF, VNQ, DBC and use either cash for the 20% portion or buy ONLY if end of month cross over 45 weeks MA, giving 12% or so continuous wining averge.
I think you may want to lok at older approach as more frequent may not be better eventhough it sound this way
|
Kevin_in_GA
4,599 posts
msg #97129
- Ignore Kevin_in_GA |
10/21/2010 11:16:27 AM
VWO has now moved below both SPY and the top scorer IWM. Its weekly TSI histogram(5,5,9) has also crossed below 0.
I'l be out of VWO tomorrow and move into IWM (since I can't trade GLD in my 401k account). I think the Histogram cross is a good signal for exit, as it can get you out with a little more profit.
This is of course running exactly counter to what I just posted about waiting until the end of the week!
++++++++
Call me fickle - I am still in VWO because I decided to wait after seeing the futures looking green. Good thing I did. The TSI Histogram for VWO is back above 0, and at the moment it is again the top ETF (even beating GLD as I write this, according to SF delayed data).
My backtest data was for end-of-week, so that's what I guess I should be sticking to.
|
Kevin_in_GA
4,599 posts
msg #97176
- Ignore Kevin_in_GA |
10/24/2010 11:21:39 AM
IWM is the new top scoring ETF - was jumping back and forth between IWM, SPY, and VWO during the middle of the week.
Truth be told, all are still looking good, but the system says to go into IWM. With the election right around the corner, who knows what will happen ...
|
davesaint86
726 posts
msg #97248
- Ignore davesaint86 |
10/29/2010 11:10:10 PM
Hi Kevin!
I noticed that ETF Replyy allows you to test Bi-Weekly now using the Relative Strength Portfolio option Also for those of you that have IRAs or non-401 K accounts the following ETFs seem to produce good results backtesting in ETFReplay. Buy the top tow ETFs using the monthly signals at 50/50 since 2003 shows a return of 663%. However in 2010 Semi-Monthly returns a better result than monthly. The list below is the list that www.decisonmoose.com basically using. I added BIV and VWO.
symlist
(BIV,VWO,EPP,EWJ,GLD,IEV,IWM,SHY,TLT,ILF,SPY)
Also, the below strategy has worked pretty well buy buying the top ETF Semi-Monthly. ETFReplay shows since 2008 a gain of 135%.
BIV,DBA,EFA,GLD,IWM,IYE,SHY,TLT,VWO
I loaded both these lists using the TSI filter you built. For my 401k I use EFA, AGG, IWM, SHY.
-Dave
|
duke56468
683 posts
msg #97326
- Ignore duke56468 |
11/4/2010 4:16:55 PM
Dave..........Could you explain your settings a little more on ETF RePlay? I didn't get even close to 663% back to 2003 with the symbols you shared. Did you use the standard 40/30/30 on the weighting? What was the 50/50 you mentioned? Thanks in advance.
|
davesaint86
726 posts
msg #97331
- Ignore davesaint86
modified |
11/4/2010 11:02:35 PM
Duke
Time Period Weight
ReturnA 3-Months - 50%
ReturnB 20-Days - 50%
Volatility 20-Days 0%
Buy top two
Monthly
Since 2003
Total Return 704%
Symbols in Portfolio - BIV,EPP,EWJ,GLD,IEV,ILF,IWM,SHY,TLT,VWO
Current Top Two - EPP, GLD
Stock Fetcher Top Two - IWM, IEV (Using Kevin's WEEKLY TSI(5,5,1)
Just a note that for 2010 - buying Monthly has not done as well as buying bi-monthly in 2010.
Dave
|
duke56468
683 posts
msg #97340
- Ignore duke56468
modified |
11/5/2010 10:30:29 AM
Thanks Dave I appreciate the response. If you use AGG in place of SHY it jumps to 747% with no loss in volatility.(2 basis points).
|