Kevin_in_GA
4,599 posts
msg #91970
- Ignore Kevin_in_GA |
5/3/2010 9:47:05 AM
Note 1: My VTI, SPY or DIA numbers do not include dividends reinvested. Do yours?
Note 2: My numbers do not includes taxes or commissions. Do yours?
Note 3: VTI, SPY and DIA numbers come from VectorVest software.
Your study honors Warren Buffet's best advice: "The best way to make money is FIRST, don't lose money." Thus, your 2008 shines relative to millions of investors who lost $$$$$$$$$$$. How many wouldn't have gladly accepted a -3.92% loss in 2008?!
Did your study include these three inverse ETFs: DOG, SH and PSQ. Or not?
+++++++++++++++++++++++
1. No.
2. No.
3. I did not include any inverse ETFs - just those shown in the symlist of the filter. Obviously you can add in any other symbols you want to to see how they would fare - the inverse ETFs would have done well during the big downturn, but would have lost you money in March 2009 and Feb 2010. You are welcome to add them in and retest to see if they help.
I have been fooling around with the ideas of relative strength and risk-adjusted returns for the last month or so, prompted by the clever work at etfreplay.com. By the way, their 50/50 selections returned around $227,000 over the same time frame.
I do not want people to think that volatility should be ignored just because the returns are higher - that is to be expected. What is more important is to be sure that you are aware of the current volatility and risk when placing the trade. This is why the Sharpe ratio is such a useful metric, and why my original filter focused on it rather than just relative return.
The relative return filter I wrote here is a quick means to the same end, but in reality I would rather use one that looked at the sum of the Sharpe ratios, and let you see the relative volatility even if you do not use it in your calculation. You might see a lower overall return, but it will also come with lower volatility and drawdown.
|
cwn6161
40 posts
msg #91971
- Ignore cwn6161 |
5/3/2010 9:49:49 AM
Kevin - you've done a great job with this. I was considering using your modified TRO crock pot, but I was worried about trading the lower priced stocks and how easily they could move, depending on the amount one was trading. This filter really is great for people with 401ks or other longer-term investment portfolios. Well done!
|
guymar
113 posts
msg #91984
- Ignore guymar
modified |
5/3/2010 12:00:16 PM
Kevin, great post, could you just tell the ignorant like me how the relative strength is computed exactly? Is it comparative relative strength like in the SF user guide? It seems very difficult to find a unique formula for it ....?
|
Kevin_in_GA
4,599 posts
msg #91986
- Ignore Kevin_in_GA |
5/3/2010 2:08:15 PM
Relative strength and comparative relative strength are the same.
As you can see from the filter above, only one column is generated.
|
sbuck143
88 posts
msg #91987
- Ignore sbuck143
modified |
5/3/2010 2:12:39 PM
=thunderous applause for Kevin=
Great job Kev! Now, to really blow ones mind, imagine buying a 6 month out call instead of the ETF each month.....
|
sbuck143
88 posts
msg #91991
- Ignore sbuck143 |
5/3/2010 4:23:39 PM
I am wondering what would be the best mixture of ETF's to use with this theory?
ETFreplay is a great site, but I wish you could easily test more than just 2 at a time (with SHY being locked in as one of them). There's a very long way to go about it by comparing and combining different sets of data, but it seems like this would work best with maybe 5 detrended ETF's, and that gets to be exponentially more work to backtest.
|
guymar
113 posts
msg #91992
- Ignore guymar |
5/3/2010 4:27:22 PM
I am very happy with this - the underlying formula could help to build something like this for Europe - no stockfetcher over there (yet?) any additional help greatly appreciated, to do this in Excel or whereever ... we are just dividing a number of closes over a number of periods here? Am I right?
|
Kevin_in_GA
4,599 posts
msg #91995
- Ignore Kevin_in_GA |
5/3/2010 5:22:43 PM
In essence, yes. The relative strength calculation is simply the return of the ETF over the last 20 or 63 days divided by the return for the SPY over the same time period.
|
sbuck143
88 posts
msg #91998
- Ignore sbuck143 |
5/3/2010 5:43:57 PM
I like your thinking Kevin in subbing out SPY for IWM.
Here would be my uber basket selection, which I will attempt to backtest as I get the time.
FXI (China)
EWW (Mexico)
IWM (US)
LSC (leveraged commodities with much lower correlation to equities)
GLD (gold)
TLT (long term treasuries as the safe haven)
Plug in FXI and EWW in the etfreplay tester for some eye-popping results.
You'd probably want to volalitity weight in some manner given these rockets....
|
ciscoslaves
5 posts
msg #92002
- Ignore ciscoslaves
modified |
5/3/2010 6:42:31 PM
Kevin,
This shorter filter is a great idea, and I tried to analyze why.
The filter more or less tries to time the market, and signals when to go to cash. I haven't done the numbers, but it seems creating any sort of mixed fund that follows the broader market would have returned similar results. I mean most stocks have returned 100%+ since the March 2009 lows.
Your filter pretty much signaled to go to cash when the worst was upon us and prevented much of the loss. I'm wondering what the results would have been if one were to buy their preferred stocks (maybe one of the best stocks from 5 different top sectors) when the filter signaled to buy IWM or EEM.
Although I guess this idea defeats the purpose of using Risk/Reward ratios
|