ricks_stocks
35 posts
msg #96068
- Ignore ricks_stocks |
9/7/2010 8:45:03 PM
"Here is the most recent iteration of the filter I am using"
Pardon my ignorance but what within your latest iteration signifies the dominant ETF.
|
dmstcp
2 posts
msg #96069
- Ignore dmstcp |
9/7/2010 10:18:19 PM
AGG still the leader, with GLD moving up - what does that tell us?
I believe that it tell's us that the filter is sorted incorrectly.
|
Radiomuse
288 posts
msg #96081
- Ignore Radiomuse |
9/8/2010 1:36:14 PM
Tough crowd! To a guy that creates so many great filters, I think we can forgive and even correct a small oversight with the coding change "sort column 3 descending".
The dominant fund is now signified by the highest TSI value. I don't really understand the calculation behind the True Strength Indicator, except I think it measures an investments performance and momentum against it's own history.
Kevin - did you manually backtest the weekly TSI strategy, or can it be done via SF, ETFreplay, or something else?
Great work with this thread - safe travels!
|
ricks_stocks
35 posts
msg #96085
- Ignore ricks_stocks |
9/8/2010 4:13:23 PM
Thanks, Radiomuse
My question was sincere and was in no way meant to be derogatory. I to appreciate the work Kevin does and look forward to his work in the future.
|
Kevin_in_GA
4,599 posts
msg #96086
- Ignore Kevin_in_GA |
9/8/2010 4:59:20 PM
AGG still the leader, with GLD moving up - what does that tell us?
I believe that it tell's us that the filter is sorted incorrectly.
+++++++++
Honestly??
The new filter is sorted correctly - in alphabetical order from A to Z. Why? So that the corresponding columns form a properly positioned correlation matrix, with each ETF self-correlation on the major diagonal.
I fully expect that anyone using this filter can manually find the highest of the 6 values. The purpose of the new filter is to maintain and update the CORRELATION MATRIX so that you can be sure that you have a properly diversified set of asset classes from which to work.
The fact that the top ETFs are GLD and AGG tell me that caution is warranted. I cannot trade GLD or VNQ in my 401k, but included them here to help others who might be using this as an investment tool. Until VWO or another equity class scores higher than AGG, I will stay happily in Fixed Income.
|
Kevin_in_GA
4,599 posts
msg #96158
- Ignore Kevin_in_GA |
9/10/2010 5:59:47 PM
This indicator says it's time to move into GLD - or Emerging Markets if you are using this in a typical 401k.
Both have done quite well of late. The time one has spent in AGG (since 5/7) has netted you 3 precent, with essentially no risk or volatility. I will be shifting my allocation to Emerging Markets on Monday in the 401k I am managing by this method.
|
campbellb75
101 posts
msg #96161
- Ignore campbellb75 |
9/10/2010 7:54:10 PM
Kevin-
Are you sure you have the correct date stating that one would have moved their assets into AGG on 5/7? According to the filter one would have been in GLD.
Weekly TSI (5,5,1)
GLD - 54.88
VNQ - 17.12
AGG - 11.70
|
Radiomuse
288 posts
msg #96167
- Ignore Radiomuse |
9/10/2010 10:23:44 PM
I'm getting the same results for 5/7 - GLD on top with a TSI of 54.88. I really like this idea though, and think I might run a strategy like this, couple of questions....
- planning to use TLT (long-term treasuries) instead of AGG. More volatility, but also low correlation with the other asset classes.
- can a strategy like this, with weekly, bi-weekly, or monthly rebalancing be backtested in Stockfetcher? I tried to simulate a weekly rebalancing by plugging in both a minimum and maximum holding period of 5 days - I can't explain exactly what happened (weird stuff), but it definitely didn't work.
- what made you switch to a weekly rebalancing instead of monthly? With the different portfolio's I ran in etfreplay, it just seemed to lower the returns and increase the volatility...
|
Kevin_in_GA
4,599 posts
msg #96170
- Ignore Kevin_in_GA |
9/11/2010 9:04:35 AM
Kevin-
Are you sure you have the correct date stating that one would have moved their assets into AGG on 5/7? According to the filter one would have been in GLD.
Weekly TSI (5,5,1)
GLD - 54.88
VNQ - 17.12
AGG - 11.70
+++++++
You're correct. If you look at the posts in this thread, you'll see that the original desgn and filter did not include VNQ or GLD. Hence, if you had been following this, you would be in AGG all that time. I have added in GLD and VNQ to further diversify this set of asset classes, but I cannot trade either of them in the account I am using for this.
|
Kevin_in_GA
4,599 posts
msg #96171
- Ignore Kevin_in_GA |
9/11/2010 9:12:18 AM
I'm getting the same results for 5/7 - GLD on top with a TSI of 54.88. I really like this idea though, and think I might run a strategy like this, couple of questions....
- planning to use TLT (long-term treasuries) instead of AGG. More volatility, but also low correlation with the other asset classes.
- can a strategy like this, with weekly, bi-weekly, or monthly rebalancing be backtested in Stockfetcher? I tried to simulate a weekly rebalancing by plugging in both a minimum and maximum holding period of 5 days - I can't explain exactly what happened (weird stuff), but it definitely didn't work.
- what made you switch to a weekly rebalancing instead of monthly? With the different portfolio's I ran in etfreplay, it just seemed to lower the returns and increase the volatility...
++++++
You cannot run this filter on a site like ETFReplay, because the RS and TSI calculations used in each are different. TSI is an intrinsic measure of strength, normalized and bounded between 100 and -100. Relative Strength is unbounded and is measured against a benchmark such as the ^SPX. Similar in concept but different in execution.
Weekly rebalancing back tested as the best when I looked at daily/weekly/monthly. For me it represents a good compromise between responsiveness and volatility.
|