davesaint86
726 posts
msg #92149
- Ignore davesaint86
modified |
5/5/2010 5:18:57 PM
Thanks Kevin for answering all of my questions.
If anyone out there needs a conservative buy & hold ETF strategy 5-8% per year these ETFs and Weightings same to do pretty well even in 2008.
BND - 25%
BSV - 25%
SHY - 25%
GLD - 25%
Dave
|
hmsb4494
81 posts
msg #92165
- Ignore hmsb4494 |
5/5/2010 8:07:37 PM
Kevin---i'm a little confused (nothing new)
on etf backtest using eem, iwm, and shy I ran the 3 month time period with 100% weight.
I got 317% since 2003
When I ran the same using the 3 month 50% weight, and another 3 month 50% weight, I GOT 448% SINCE 2003.
Whats happening here???
|
Kevin_in_GA
4,599 posts
msg #92167
- Ignore Kevin_in_GA |
5/5/2010 8:22:06 PM
I just ran it with those settings and got the same answer for 50/50 or 100/0 = 448%. No apparent glitch.
|
hmsb4494
81 posts
msg #92168
- Ignore hmsb4494 |
5/5/2010 8:37:29 PM
try this Kevin-----3 month 100% 6 month 0
|
hmsb4494
81 posts
msg #92169
- Ignore hmsb4494 |
5/5/2010 8:39:48 PM
if you change return b time period, it changes total return---even though the weight on the different time period is 0
|
Kevin_in_GA
4,599 posts
msg #92170
- Ignore Kevin_in_GA |
5/5/2010 8:42:59 PM
Yeah, numbers are different. Hey, it's not my site, guys! I think it must be looking for some second timeframe, and factor it in even if it is set to 0.
I still think the 50/50 split numbers are solid. I also know that I backtested my shorter filter by hand from 2007 - today, and it gave very simliar results.
|
hmsb4494
81 posts
msg #92171
- Ignore hmsb4494 |
5/5/2010 8:46:57 PM
Just making sure it wasn't me.
I really appreciate your work here Kevin!
|
hmsb4494
81 posts
msg #92189
- Ignore hmsb4494 |
5/5/2010 10:48:32 PM
The following is from etfreplay support:
think about the length of the ReturnA, ReturnB and volatility time period parameters and how that may relate to
1) the inception date of the ETFs
2) when the backtest begins
you can see that there are multiple variables at work here --- each parameter you set in those drop-menus is being converted to very literal software code. if you set ReturnB to 6 months, that parameter will logically tell the computer to include only ETF's with a minimum of 6 months history. if you now set ReturnB to 12 months, that parameter will logically tell the computer to include only ETF's with a minimum of 12 months of history, even if set to a weight of zero. so it may look like you are doing the exact same thing -- but the code actually has rules based on the parameters you set. if you set a parameter differently, different code will run -- so then you have to think about why that may have occured. Many ETF's don't have long histories and so the user will at times have some trouble seeing why something like this is different -- all we can do is try to deal with any anomalies as they come up.
|
Kevin_in_GA
4,599 posts
msg #92219
- Ignore Kevin_in_GA
modified |
5/6/2010 12:41:25 PM
Their response seems reasonable. Their tool is just that - a tool to test strategies. Now that you are aware of its glitches, you can still use it very effectively. The site is a real jewel for many investors - I think it helps to put ETFs and risk/return strategies in a clear light.
I know that several posters in this thread expressed concern about buying IWM at the end of last month, given that a major correction might be beginning. In order to take advantage of this system, you will at some point need to commit and jump in (and then ignore that investment til the end of the month).
If you bought in at the close on Friday, you were in at $71.65. As of this writing, IWM is at 69.14, a drop of only 3.5%. Given that you were in IWM at the beginning of Jan at $63.19, you are still beating the market (you are up 9.4%, compared to 3.0% for the SPY).
Where will IWM be in a month? Not sure, but you are still going to be ahead of the market when we get there. If SHY rises to the top because we have had a major corrective move, then you sell IWM and buy SHY. This is a system, and like any system it will have down months - the goal is to have more up months than down, and to have more months that beat the SPY return than lose to it.
|
sbuck143
88 posts
msg #92222
- Ignore sbuck143 |
5/6/2010 1:05:54 PM
One thing the site has raised my awareness of (but with very little way to automatically test it) is how a monthly rebalance perspective would assist many other strategies.
Their MA crossover tool for example, shows the benefits of basically just checking the MA vs price one time a month and making your decision based on that. Using a 200 day SMA, checking at the end of the month eliminated the majority of whipsaws, and skyrocketed the overall return.
One thing I am not certain of on that particular aspect of their site (MA test) is that they use their "total return" chart which they tout as the only real way to look at what an ETF/stock gives you. I agree to a certain extent.....when calculating returns. But for price action, dividend payouts for example are already factored into the chart. A stock drops by the amount of the dividend, for example.
But they aren't very clear what they use for their cross-overs. I guess it really doesn't matter, but to me MA crossovers should be done with pure price action, not with something that doesn't represent the stock actually hitting a definitive price point.
|