iShares Lehman Aggregate Bond (AGG)

All Comments on AGG

  • commenter
    Geoff
    Considine
    Aug 15 01:13 PM
    My Website
    Defining a Set of Core Asset Classes [view article]
    Aquater: not sure I understand the question.

    Geoff     Reply
  • commenter
    Geoff
    Considine
    Aug 15 01:13 PM
    My Website
    Defining a Set of Core Asset Classes [view article]
    Dumbo:

    * three years for correlations -- there is a literature on the stability of correlations and other second order moments. 36 points is a nice compromise with stable stats.

    * returns are total returns-dividends reinvested
    * returns are what you are calling Rate of Change -- i.e. basic percentage change in monthly closing prices.
    Reply
  • commenter
    Aquater
    Aug 15 12:18 PM
    Defining a Set of Core Asset Classes [view article]
    Thanks for your response. I understand the rationale for three-year look-back, tentative though it is it is back-tested by you. But on what basis is the QPP projection for one year based? I have hard time finding this. All I find is the details of its being tested. I need to see some articulation about the basis and method as to how the projection is made. Thank you, again for sharing what you have. Reply
  • commenter
    dumbo
    Aug 15 10:14 AM
    Defining a Set of Core Asset Classes [view article]
    [Repeat post; no reply to yesterday's post]
    Geoff,
    Can you clarify:
    - What time frame did you use for Correlation measurement? And the pros/cons of using this time frame vs some other (e.g. why 36 months instead of 12)
    - Did you use returns based on 'raw' prices or 'dividend-adjusted' prices?
    - For returns, did you use RateOfChange (%) or log-normal returns?
    Thanks in advance.     Reply
  • commenter
    Flav
    Aug 14 09:28 PM
    Defining a Set of Core Asset Classes [view article]
    Hello Geoff:

    Thank you very much for your quick response. It does clarify.
    I'm an economics and finance student, and i often have discussions with classmates and friends about several subjects regarding portfolio management. I must say that i have used some of your articles to support some of my arguments. I believe that for long term planning, there's nothing better than the analysis being made in your articles. But suppose that some investor or trader has a shorter time horizon; that person is willing to take on larger risks, in order to achieve larger returns. Would you still recommend this individual to diversify? Or given the shorter time frame and less risk aversion, should this investor have a more concentrated portfolio?

    Again, thank you very much

    Looking forward to read more of your articles     Reply
  • commenter
    Geoff
    Considine
    Aug 14 08:07 PM
    My Website
    Defining a Set of Core Asset Classes [view article]
    Flav:

    I am not saying that historical information is of no use--rather than that data must be tempered with a model that reduces your reliance on the specific relative performance of those assets in the historical period selected. The challenge with how much data to use is the motivation for forward looking models like QPP. If you use too long an historical data record, you will end up with stats that do not reflect the evolution of the markets. How can you model tech stocks or internet stocks if you want to use 40 years of data, for example? The domestic auto industry is in an entirely different state than it was even 15 years ago...

    I have tested and benchmarked QPP using three years of data to initialize the model and then QPP brings in long-term analytics on the way risk and return balance in capital markets--i.e. you don't need security specific data for this. RiskMetrics is even more extreme in terms of emphasis on recent data--they use an exponentially decaying weighted average going back in time for many parameters.

    I do not know that 3 years, my default in testing, is optimal--but it has tested out well for long-term tests spanning decades (I have articles on this with out-of-sample tests).

    Does this clarify?

    Geoff     Reply
  • commenter
    Flav
    Aug 14 07:48 PM
    Defining a Set of Core Asset Classes [view article]
    Hi Geoff. I made a question to you in other article that wasn't answered, so i will reformulate it in this one.
    You say that hisorical data is of no use, but in order to obtain forward looking assets class combinations with QPP, you have to enter past information about those assets classes. So, as i asked before, to determine the correct weight asigned to each holding (the way that produces the highest return with a given level of risk), should the oldest available information be introduced? or is it sufficient with just a few years back?     Reply
  • commenter
    dumbo
    Aug 14 03:56 PM
    Defining a Set of Core Asset Classes [view article]
    GC, Cna you calrify:
    - What time frame did you use for Correlation measurement? And the pros/cons of using this time frame vs some other.
    - Did you use returns based on 'raw' prices or 'dividend-adjusted' prices?
    - For returns, did you use RateOfChange (%) or log-normal returns?
    Thanks in advance.     Reply
  • commenter
    Geoff
    Considine
    Aug 14 02:18 PM
    My Website
    Defining a Set of Core Asset Classes [view article]
    Aquater: QPP is available to retail investors and advisors alike--see quantext.com. As far as how it works, I have more than 1000 pages of documentation and tests on my website (free) and much if it is also published here on SA. I suggest that you may want to start with this article:

    seekingalpha.com/artic...

    History has a place in analysis but this must be conditioned with models that capture the fact that just because an asset class out-performed in your specific historical sample, it may not out-perform in the next. Pure historical analysis gets tripped up by this.

    Part of what I am trying to explain is that institutional investors manage portfolios very differently than retail investors. It is not just about a few gurus. Again--this is discussed in many of my articles.

    Thanks for the comments.

    Geoff     Reply
  • commenter
    Aquater
    Aug 14 01:52 PM
    Defining a Set of Core Asset Classes [view article]
    This is a very useful article. It is thoughtful and provokes thinking. A great piece. I have following points, however, to further the dialog.
    1. On the one hand you say timing cannot be done and so one must stay invested in multiple widely uncorrelated asset classes. On the other hand, you want to look forward and marginalize history as backward look. Trying to have it both ways?
    2. Also, no clue is provided as to how exactly you look forward. What tools are used to look forward and how are they better than other timing tools? Just saying QPP is not enough. You may be holding QPP as a proprietary device, but you would gain credibility if you share at least some philosophy behind it.
    3. Thinking about reader comments, the name-calling match between retail and elite was entertaining heat but did not provide much light. The guru stuff was highly speculative. Both timers and non-timers have to resort to past record to justify themselves. It is indeed hard to really discount history. Snap discrediting of it as nothing but looking backward is a bit dogmatic. Its usefulness may be limited but it surely is debatable how to assess it.
    4. It is not consistent to talk about cash as wasting asset because of inflation. To be consistent all assets and not just cash should be systematically weighed against inflation. Any way, cash is better than losing your capital. Including it as one of so many others is not a sin at all. It is also important to take advantage of emerging opportunities. It would be simpler and easier to do it if you have cash.
    5. You hint at percentage allotment to the asset classes. Would like to know more about the concepts behind this.
    All in all, your article is highly beneficial, thoughtful and noteworthy.

    Reply
  • commenter
    granger
    Aug 14 01:39 PM
    Defining a Set of Core Asset Classes [view article]
    Great info as usual. I always look forward to your articles. I have to read them a couple of times (my way of understanding things), but always worth the effort. Thank You Reply
  • commenter
    Geoff
    Considine
    Aug 13 10:39 PM
    My Website
    Defining a Set of Core Asset Classes [view article]
    ZPTDO: Uhh, I think you missed the point. Your link is to inverse ETF's and yes, if you combine them with long ETF's, you end up with almost zero net return and zero risk. That is not at all what we are discussing here. These are all LONG ETF's. Reply
  • commenter
    zptdo
    Aug 13 09:45 PM
    Defining a Set of Core Asset Classes [view article]
    "I show an All-ETF portfolio that is down only 0.5% in the last year or so. " By design ETFs as a basket will cancel each other. All-ETF will show very little change at any time but will make your broker very happy :)).
    www.proshares.com/abtf...     Reply
  • commenter
    BlueDog
    Aug 13 08:02 PM
    My Website
    Defining a Set of Core Asset Classes [view article]
    Nice article, Geoff. The correlation between Vanguard Large Cap Stock Fund [VLACX] vs. Mid Cap Stock Fund [VIMSX] vs. Small Cap Stock Fund [NAESX] was particularly interesting. In defining a set of core asset classes, I think it's also important to keep expenses as low as possible in order to maximize performance. I've created a core portfolio which creates diversification at a very low expense ratio/cost: EEM, EFA, IWM, IVV, SHY, TLT, LQD, TIP, RWR. I'll have to look at DJP and IGE as potential adds. Thanks for the article.

    BD     Reply
  • commenter
    Geoff
    Considine
    Aug 13 06:09 PM
    My Website
    Defining a Set of Core Asset Classes [view article]
    FYI--a useful article on this theme:

    online.barrons.com/art...

    Reply
Loading...
Symbols:
Authors:
Quick Links: