StockFetcher Forums · Filter Exchange · PORTFOLIO SELECTION AND MANAGEMENT USING RISK/REWARD RATIOS<< 1 ... 14 15 16 17 18 ... 65 >>Post Follow-up
14 posts
msg #94469
Ignore budandann
7/1/2010 12:22:37 PM

...this is without question the "greatest" ... thread ... EVER created on SF !!!

Thank you so much Kevin (& other helpful contributors) for sharing all the good stuff.

I would like to respectfully suggest that the "AMBIEN" poster just go ahead and go back to sleep and leave this thread to the rest of us who appreciate all the good information.

2,025 posts
msg #94475
Ignore alf44
7/1/2010 3:02:08 PM


14 posts
msg #94485
Ignore budandann
7/1/2010 8:14:54 PM

thanks alf ... sweet dreams

74 posts
msg #94517
Ignore jnafach
7/3/2010 1:35:59 AM

I tried same selectons an dnot getting 877% but 530% , am I doing something wrong

40 posts
msg #94524
Ignore cwn6161
7/3/2010 11:55:15 AM

- Ignore jnafach 7/3/2010 1:35:59 AM

I tried same selectons an dnot getting 877% but 530% , am I doing something wrong


What selections are you talking about? Are you backtesting on SF or ETF replay?

The differences between the backtesting of these two programs has been discussed earlier in the thread.

14 posts
msg #94531
Ignore budandann
7/3/2010 6:54:42 PM

- Ignore gmoney
modified 6/30/2010 4:47:10 PM

Buy Top: 1 ETF
Benchmark: SPY
Update Schedule: Monthly
....................Time Period......Weight
ReturnA..... 10-Days............. 50%
ReturnB..... 20-Days............. 30%
Volatility..... 20-Days............. 20%
Backtest Start: .....2003

This backtest provided an astounding Total Return% of 867.2% vs 36.5% total return% from SPY (benchmark) from 2003 through 29 June 2010.
If I run the ETF replay backtest with the exact same settings from above today I get 513.5% vs 33.9% for SPY. I used EEM,EUO,FXE,IWM,RWM,SH,SPY,UDN,UUP for the ETF list. I copied & pasted the ETF list so there were no mistypes.

8 posts
msg #94539
Ignore gmoney
7/4/2010 4:00:21 AM

Checked the latest postings late Saturday night and learned of this disconnect.

I ran the exact same settings and I now get the same result as you. I was equally puzzled by the 513% value now being returned. I have a hard copy printout with both graphs and monthly incremental data showing the 867.2% Return with 27.0% volatility (and several other earlier printouts with progressively higher returns in the 600% - 700% - 800% + range, as I changed each of the variables to increase the returns). How could this be? I also wondered if any other SF members printed or viewed the 867% returns and incremental portfolio backtest data prior to 1 July as I had?

I quickly compared the model settings on the 867% printout with the printout showing 513% returns and monthly ETFs and monthly performance data. Both show the exact same historical ETF Buys each month and the exact same % Returns & % Volatility values each month going back to 2003, except for Nov '08 - Dec '08 and Dec '08 - Jan '09:

On the 867% data, Nov-Dec showed RWM at +23.34%, +0.98%, 139.9% and 46.0%.
On the 513% data, Nov-Dec showed RWM at -8.71%, +0.98%, 48.5% and 38.7%.

On the 867% data, Dec-Jan showed RWM at +8.53%, -8.21%, 48.4% and 38.7%
On the 513% data, Dec-Jan showed IWM at -9.67%, -8.21%, 48.5% and 38.7%

Using the exact same model settings in both runs, the report printed after 30 June should have produced exactly the same 867% results, revised upward only by the 1-2 July returns. However, these two months now show significantly different (lower) return % values that may correlate to, when re-compounded, the lower cumulative total return of 513%.

Sunday Update at 10:45 AM:

I am speculating now, but perhaps that on or about 1 July, after ETFReplay's June month end number crunch, they elected to finally back-adjust these two index ETFs in their Nov '08- Jan '09 time period to account for the unusual distribution event experienced in December '08 on ProShares ETFs, including RWM & SH - see below:

(ETFReplay provided me the following link and explanation in response to other questions I had. I had sent them the list of the specific ETF's in the test portfolio, with all model settings and the huge return it provided as the basis for my questions.

When I read this email a couple of days ago regarding distribution anomalies, I assumed the events that occurred back in 2008 were already incorporated and reflected into their 2008 historical data. It now seems not so on the 867% data, and appears that the actual Dec '08 ETF distribution changes to their database weren't incorporated until after 30 June month-end for some reason (to be verified by ETFReplay).

ETFReplay email response received a few days ago responding to different questions, but included the following:

"If you refer to the ProShares website for their cash distributions (see attachment with website address) -- you will see something very unusual occured in December 2008 --- ProShares paid out nearly $25.00 per share on RWM due to some accounting peculiarities of their holdings. This gets very complicated very quickly and you should research it yourself -- but basically, if you were long RWM on the distribution ex-date you were paid $24.57 in cash into your account and the ETF price adjusted down by that amount overnight. This is effectively a dividend payment, not a loss.

So the question is over how to account for this fact as you did not lose the $24.57 outright --- you just were paid it in cash. If you calculate the price changes on Stockcharts -- then the calculation will be WAY off as you need to somehow account for that $24.57 that you earned in cash. Stockcharts simply ignores it -- as does Google and Yahoo and Ignoring it doesn't address the situation correctly.

Now, this is a very unusual situation that was written about in-depth at the time -- here is one article:

Quotes from this article - "Many shareholders in the ProShares ETFs were surprised on December 23 when the funds traded down sharply, but that is simply how distributions work. For instance, for SIJ, ProShares made a distribution of $47.85/share on December 22; as a result, the fund opened down $47.85/share on the morning of December 23.
That money doesn't simply disappear. Rather, shareholders who held SIJ on December 22 will receive a cash payment of $47.85/share in their brokerage accounts by the end of this year."

ETFReplay continues, "So basically, our (ETFReplay) software needs to make an assumption -- and our assumption is consistent with how indexes are calculated -- we assume you re-invest the dividend right away. This isn't perfect -- but what it DOES do is it allows you to understand what the underlying indexes are doing correctly as this method is entirely consistent with how you calculate the total return of an index -- while Stockcharts is not.

SH was also affected during this time as seen in the table above.

So the bottom-line is that we (ETFReplay) offer our software as a guideline only. Users should verify results for themselves. If you would like to learn how to calculate this and why we do it the way we do it -- you should research 'total return' and find examples of comparing 'total return' to 'price return' -- and then do a few calculations and see for yourself how it works." End of quote.

So, I question why they elect to make a change at the June 2010 month-end update to historical data on these two ETFs back in the 2008 time period instead of back in 2008? I will ask ETFReplay to comment on that when they return from holiday.

Tomorrow, I'll load both sets of historical data on a spreadsheet and see if I can duplicate the "before & after" on both Total Return runs and to "re-invest" those specific dividends right away, in hopes that the Total Return % will reconcile back to the 867% compounded value.

Again, the point of logic here is by using the exact same model settings & the exact same ETFs, how likely is it to experience significant historical differences NOW in June 2010, instead of when they knew about it in December 2008? If they did change their approach on how it is shown, why did they change it now?

Before this discovery I sent them another email on Friday, to get from them a clear understanding of every action taken by the model, and when each action must have been taken each month, in order to have duplicated the past performance results of this portfolio since 2003. In order to actually trade based on this tool, we'll need know the what and when to do the same things going forward in managing a funded model. Awaiting a response on that email. When I hear back, I will share their response.

Enjoy a Safe and Patriotic July 4th Independence Day Weekend.

683 posts
msg #95198
Ignore duke56468
7/29/2010 9:29:35 AM

The VNQ-EEM combination on etfreplay back tested well (211%) since 2005. Do you think it makes sense to add VNQ to the symlist on this filter?

3,829 posts
msg #95199
Ignore Kevin_in_GA
7/29/2010 9:47:38 AM

The VNQ-EEM combination on etfreplay back tested well (211%) since 2005. Do you think it makes sense to add VNQ to the symlist on this filter?

Given that it is a different asset class, from a diversification standpoint it makes perfect sense.

You want to be sure that you portfolio of ETFs is not too highly correlated - if so they will all move in the same direction and it will be harder to minimize volatility. You can go to can determine the correlation between any two ETFs over different time periods.

VNQ- SPY correlation for the past 6 months: 0.94

VNQ- EEM correlation for the past 6 months: 0.87

VNQ- IWM correlation for the past 6 months: 0.94

VNQ- GLD correlation for the past 6 months: 0.00

VNQ- SHY correlation for the past 6 months: -0.67

VNQ- BND correlation for the past 6 months: -0.64

142 posts
msg #95872
Ignore davesaint86
8/28/2010 6:52:43 PM

Is anyone still using this strategy in their 401k's or IRA's? If not, why not?



StockFetcher Forums · Filter Exchange · PORTFOLIO SELECTION AND MANAGEMENT USING RISK/REWARD RATIOS<< 1 ... 14 15 16 17 18 ... 65 >>Post Follow-up

*** Disclaimer *** does not endorse or suggest any of the securities which are returned in any of the searches or filters. They are provided purely for informational and research purposes. does not recommend particular securities., Vestyl Software, L.L.C. and involved content providers shall not be liable for any errors or delays in the content, or for any actions taken based on the content.

Copyright 2016 - Vestyl Software L.L.C.Terms of Service | License | Questions or comments? Contact Us
EOD Data sources: DDFPlus & CSI Data Quotes delayed during active market hours. Delay times are at least 15 mins for NASDAQ, 20 mins for NYSE and Amex. Delayed intraday data provided by DDFPlus