Seeking Alpha
View as an RSS Feed

Lowell Herr  

View Lowell Herr's Comments BY TICKER:
Latest  |  Highest rated
  • Faber 10 Portfolio: Combining Low Correlated ETFs With Momentum Model [View article]
    User,

    I hope your read my article on back-testing.

    http://seekingalpha.co...

    One of the most tested portfolios can be found on my blog and it is the Rutherford Portfolio. Not all blogs related to this portfolio are available without a Platinum membership.

    Lowell
    Apr 14, 2015. 01:01 PM | Likes Like |Link to Comment
  • Faber 10 Portfolio: Combining Low Correlated ETFs With Momentum Model [View article]
    Noumann,

    I am using 182 calendar days for the correlation analysis look-back.

    Lowell
    http://itawealth.com
    Apr 14, 2015. 11:45 AM | Likes Like |Link to Comment
  • Back-Testing Results Are Filled With Uncertainty - Turn On Your 'Doubt Meter' [View article]
    TMD,

    I agree with your points above and here is how I attempt to "cure" the problems. My "solutions" match the answers you have provided.

    1. The 33-day portfolio reviews also include running a fresh correlation analysis as well as ETF rankings. This is the dynamic (adaptive) model of which you speak.

    2. You answered the second problem as well. In addition to using SHY as the cutoff or "circuit breaker" ETF when a 2008 bear market strikes and all securities become highly correlated, the ETF ranking spreadsheet (not pictured) also includes addition warning triggers of when to exit the market. Whipsaws continue to be a nagging problem in flat markets. Waiting 33 days to update a portfolio helps to reduce the impact of whipsaws.

    3. To handle the problem of both securities being highly correlated and going in the wrong direction, I use more than three or four ETFs as potential building block candidates. How I attempt to manage this might be the theme of another article.

    Thanks for your comments as they are spot on.

    Lowell
    http://itawealth.com
    Apr 14, 2015. 06:39 AM | Likes Like |Link to Comment
  • Back-Testing Results Are Filled With Uncertainty - Turn On Your 'Doubt Meter' [View article]
    Fred,

    I show in the above article that TLT made a significant (positive) impact on the returns. That may not be repeated in the future. At least I would not count on it.

    There are several factors that color back-testing results. 1) Choice of securities. 2) Start and stop dates. 3) Look-back period. 4) Number of days between portfolio review. These are four biggies.

    Lowell
    http://itawealth.com
    Apr 13, 2015. 04:30 PM | 1 Like Like |Link to Comment
  • Back-Testing Results Are Filled With Uncertainty - Turn On Your 'Doubt Meter' [View article]
    Pete,

    What you suggest is partially what I'm doing on my blog. 1) I'm tracking 14 portfolios with the review periods occurring every 33 days so the portfolio updates are scattered throughout the month. 2) While there is a lot of overlap when it comes to the ETFs used to populate the portfolios, there are variations. 3) There are slight variations in how the momentum model is applied. 4) Portfolio IRR values are reported each week along with performance trends compared to what was happening six weeks ago and one year ago.

    Yes, it takes a lot of effort to keep all this up, but that is what the blog is all about. If you are interested in a sample portfolio (not a real one on my blog) check out this link.

    http://seekingalpha.co...

    Lowell
    http://itawealth.com
    Apr 13, 2015. 04:10 PM | 1 Like Like |Link to Comment
  • Back-Testing Results Are Filled With Uncertainty - Turn On Your 'Doubt Meter' [View article]
    Pete,

    " There's clearly a significant amount of excess return to be gained (or lost) by any a priori choice of review dates."

    Correct. Uncertainty of return is what back-testing is showing at this end. This is one of the uncertainties listed above and why readers need to be skeptical when they see a specific single return result from one back-test.

    Lowell
    http://itawealth.com
    Apr 12, 2015. 05:11 PM | Likes Like |Link to Comment
  • Back-Testing Results Are Filled With Uncertainty - Turn On Your 'Doubt Meter' [View article]
    Varan,

    One of my next projects is to take the commission free ETFs provided through TD Ameritrade and run correlation analysis until I cull the number to something under 15. I suspect the number will come in under 10 as I know right out of the corral many are highly correlated. Once I have low correlated ETFs isolated, then run a similar analysis as was performed on the ETFs for this article. I expect similar winning results.

    Lowell
    http://itawealth.com
    Apr 12, 2015. 03:40 PM | Likes Like |Link to Comment
  • Back-Testing Results Are Filled With Uncertainty - Turn On Your 'Doubt Meter' [View article]
    Hi Varan,

    Point #1 above: I totally agree with your point. To carry this argument one step further, a portfolio benchmark should also be available for investing.

    Point #2: SHY is included as a low volatile ETF in an effort to keep one out of major bear markets. As a practical matter, I have yet to find myself in a situation where I was fully invested in SHY.

    An investor has a high probability of achieving positive returns by using VTI, VEA, and a bond ETF such as TLT, BIV, or BND to construct their portfolio.

    Point #3: The more I read from the Dual Momentum book the less I see that is all that different from articles that go under the name of "Adaptive Asset Allocation." It seems as if every author is seeking a special title to distinguish similar momentum models.

    Lowell
    http://itawealth.com
    Apr 12, 2015. 11:27 AM | Likes Like |Link to Comment
  • Back-Testing Results Are Filled With Uncertainty - Turn On Your 'Doubt Meter' [View article]
    LHZ,

    The ETFs used in the above article mirror the GEM model, but obviously are different. I chose to use commission free ETFs (through TD Ameritrade) as those are ETFs I use with portfolios I track. The three basic asset classes Antonacci uses in his book are covered in the two examples provided in my article.

    Using my ranking model, an investor would now be invested in EFA. I used EFA instead of VEA as it has a longer historical record. Currently, I am invested in VEA as it is the highest ranking ETF of the three.

    I'm still confused why Antonacci used ACWI and then called for an international ex-US ETF. I thought he would use ACWX as that is the ex-US international ETF.

    The intent of my article is not to challenge the GEM model. Rather, I wanted to point out inherent uncertainties that surround all back-testing models. These uncertainties also apply to back-tests I use to justify investing models I use so I am not immune to the problems.

    Lowell
    http://itawealth.com
    Apr 12, 2015. 11:14 AM | Likes Like |Link to Comment
  • Back-Testing Results Are Filled With Uncertainty - Turn On Your 'Doubt Meter' [View article]
    Hsfrey,

    There was one starting date (June 30, 2006), not 33. Thirty-three (33) days is the review interval, but this sometimes falls on a non-business day. The fellow who ran these back-tests (programmed in R) has a routine that finds the nearest trading day to the calculated check date. When I am running portfolio reviews in real time, if the review date falls on Saturday, Sunday, or a holiday, I shift the review date to the first trading day following the "off" day.

    The following quote comes from the programmer. "At each calculated check point a random number of trading days are added by selecting a number between 0 and 14 trading days from a flat random number generator. Each 8 year run began on June 30, 2006, so no random days were added to the starting date."

    I hope this helps to answer your questions.

    Lowell
    http://itawealth.com
    Apr 11, 2015. 08:53 PM | Likes Like |Link to Comment
  • Back-Testing Results Are Filled With Uncertainty - Turn On Your 'Doubt Meter' [View article]
    Govind Das,

    I'm aware of many, not all, of the back-testing results. However, I am using different arrays of securities in different portfolios that were launched at different times. In addition, one of my benchmarks is actually customized for each portfolio. In other words, the variables are sufficiently different that it requires me to prove to my own satisfaction whether or not the momentum anomaly holds going forward.

    Lowell
    Apr 11, 2015. 04:28 PM | Likes Like |Link to Comment
  • Back-Testing Results Are Filled With Uncertainty - Turn On Your 'Doubt Meter' [View article]
    Govind Das,

    I should have made a clearer distinction between the basic uncertainties inherent in back-testing vs. momentum model results. I'm not faulting the momentum model as I think it has merit.

    As for the 8-year study vs. a 40-year study, I did not have 40 years of historical data for any of the ETFs used in the two examples. It is unclear to me how Antonacci was able to come up with results beginning in 1974 (Table 8.4 on page 102) where he uses ACWI and AGG as two ETFs. ACWI, according to Yahoo-Finance, began serving up data on March 28, 2008 and AGG data begins on September 29, 2003. What proxies were used for aggregate bonds and ex-U.S. Equities to produce 1974 - 2013 results.

    On page 101, ACWI is the ETF used to represent ex-U.S. Equities. Did Antonacci really mean to use ACWI, not ACWX? There is confusion at the very least as one ETF includes U.S. stocks while the other is ex-US stocks.

    Another problem I have with the data tables throughout the book has to do with pushing the data to four significant figures. It is absurd to think the return percentages are accurate to the hundredth place. It is this type of reporting one frequently finds in investment books that do not pass the smell test. Even if it were actual data, reducing accuracy to one decimal is plenty.

    I hope these additional comments shed additional light on why I light up my "doubt meter" when examining any type of back-testing or future projections. I'm sure I too have been guilty of taking computer generated numbers and presenting them as "truth."

    Respectively,

    Lowell
    http://itawealth.com
    Apr 11, 2015. 02:37 PM | Likes Like |Link to Comment
  • SYLD: An ETF You Can Buy And Hold Forever [View article]
    I ran a comparison between SYLD and VTI and see they are highly correlated. SYLD costs 59 basis points while VTI is only 5 basis points. Over a long period that difference will add up to real dollars.

    My preference is to stick with the lower cost ETF, VTI.

    Lowell
    http://itawealth.com
    Apr 11, 2015. 12:46 PM | 8 Likes Like |Link to Comment
  • Back-Testing Results Are Filled With Uncertainty - Turn On Your 'Doubt Meter' [View article]
    Richard,

    The two portfolio models shown in this article were reviewed every 33 days so the draw-downs or portfolio reversals are based on the time of review. What happens between the review periods is neglected.

    The primary reason for the 33 day review interval is to shift the portfolio update throughout the month. Of the 14 portfolios I am tracking, all are reviewed on a 33 day interval. After a review the portfolio goes into "neglect" mode for another 33 days.

    Lowell
    http://itawealth.com
    Apr 11, 2015. 10:53 AM | 1 Like Like |Link to Comment
  • Back-Testing Results Are Filled With Uncertainty - Turn On Your 'Doubt Meter' [View article]
    Eric,

    Yes, I think there is validity to the momentum approach (momentum anomaly) used to generate the above results. Nevertheless, I want to test the model using out-of-sample data and that is what I am attempting with several portfolios.

    I'm also comparing momentum models vs. passively managed portfolios to see if there are major differences that come about with extra work.

    Lowell
    http://itawealth.com
    Apr 10, 2015. 08:20 PM | 1 Like Like |Link to Comment
COMMENTS STATS
1,119 Comments
335 Likes