There are very cool tools offered in this answer here. One of them is this here that allows you to see SD and return in rebalanced portfolio over longer time periods. Now I am trying to compare the results to portfolios that have no rebalancing. How can I do that?

The reason for the"historical" effect is the long-term. I am not interested in short-term models such as brownian stochastic process. I hope you understand that it implies challenges of its own, not just calculations.

  • Hey, if that's so, where's my "accepted answer"? MBhunter and I are in a race to the top. Well, he's there, but I'm on his tail. Only one poster can be number one. (by the way, glad you liked my find) – JoeTaxpayer Jun 28 '11 at 20:08
  • @Joe: Oh, I see. It's all about the reputation for you, isn't it? :) – mbhunter Jun 28 '11 at 20:50
  • @mbhunter - shallow of me? :P – JoeTaxpayer Jun 29 '11 at 0:10
up vote 3 down vote accepted

Do not reinvent the wheel!

Historical data about stock market returns and standard deviations suffer from number of issues such as past-filling and mostly survivorship bias -- that the current answers do not consider at all. I suggest to read the paper "A Century of Global Stock Markets" by Philippe Jorion (UC Irvine) and William Goetzmann (Yale), here. William Bernstein comments the results here, notice that rebalancing is sometimes a good option but not always, his non-obvious finding where the low SD did not favour from rebalancing:

Non-obvious example, you can use mean variance optimizer to investigate this kind of situations, source EfficientFrontier.com -article above.

Look at the final page of the paper, "geometric returns -- represent returns to a buy-and-hold strategy" and the "arithmetic averages -- give equal weight to each observation interval.", where you can find your asked "historical effect of Rebalancing on Return and Standard Deviation". The paper nicely summarizes the results to this table:

Historical returns and Standard deviations, source the paper above.

The results in the table are from the interval 1921-1996, it is not that long-time but even longer term data has its own drawbacks. The starting year 1921 is interesting choice because it is around the times of social-economical changes and depressing moments, historical context can be realized from books such as Grapes Of Wrath (short summary here, although fiction to some extent, it has some resonance to the history). The authors have had to ignore some years because of different reasons such as political unrest and wars.

Instead of delving into marketed spam as suggested by one reply, I would look into this search here. Look at the number of references and the related papers to judge their value.

P.s. I encourage people to attack my open question here, hope we can solve it!

  • Doesn't the arithmetic average in the chart above presume re-balancing on a strictly time based ('observation interval') method? Considering that is only one proposed basis for re-balancing is it fair to use such a narrow approach as a broad brush to paint all re-balancing as flawed or ineffective? (here's an example of another theory: iijournals.com/doi/abs/10.3905/jpm.2003.319894 which I don't think can be modeled by using a simple arithmetic average based on equally weighted observation intervals ) – Chuck van der Linden Jul 1 '11 at 7:04
  • @Chuck van der Linden: the author is answering this question with the graph "Does it pay to rebalance assets with a 1% return difference compounded over 76 years?"´, assuming "a 1% domestic stock advantage, a 0.5 correlation between US and foreign markets, and a zero return/zero SD for t-bills"´. Sorry cannot get access to your article now but as far I can understand the author's example is ad hoc by nature. I am not actually sure whether the graph has a mistake with its labels. Why does it have geometric and not arithmetic there? He explains like comparing rebalancing and no-rebalancing. – user1770 Jul 2 '11 at 19:16

Doesn't "no rebalancing" mean "start with a portfolio and let it fly?" Seems like incorporation of rebalancing is more sophisticated than not.

Just "buy" your portfolio at the start and see where it ends up with no buying/selling, as compared with where it ends up if you do rebalance.

Or is it not that simple?

  • 1
    +1 - sometimes simple is right. This would be my answer as well. Right to the point. – JoeTaxpayer Jun 29 '11 at 0:57
  • -1 not answering the question – user1770 Jun 30 '11 at 3:25
  • 1
    @hhh: How did I not answer the question? Sounds like the OP already has a way to get the rebalanced amount. That's the harder calculation. – mbhunter Jun 30 '11 at 5:10
  • @mbhunter: it is more a good comment rather than a good answer. It does not attack the question in the title, it is more rhetoric and hence better for comment. This question do have scientific answers if you really look hard enough missing things such as confidence intervals and critique about "historical data" itself that can be tampered and corrupted. – user1770 Jun 30 '11 at 14:02
  • 1
    "Now I am trying to compare the results to portfolios that have no rebalancing. How can I do that?" That's the question I was answering. – mbhunter Jun 30 '11 at 16:46

To answer your question directly.. you can investigate by using google or other means to look up research done in this area. There's been a bunch of it

Here's an example of search terms that returns a wealth of information. effect+of+periodic+rebalancing+on+portfolio+return

I'd especially look for stuff that appears to be academic papers etc, and then raid the 'references' section of those. Look for stuff published in industry journals such as "Journal of Portfolio Management" as an example.

If you want to try out different models yourself and see what works and what doesn't, this Monte Carlo Simulator might be something you would find useful

The basic theory for those that don't know is that various parts of a larger market do not usually move in perfect lockstep, but go through cycles.. one year tech might be hot, the next year it's healthcare. Or for an international portfolio, one year korea might be doing fantastic only to slow down and have another country perform better the next year.

So the idea of re-balancing is that since these things tend to be cyclic, you can get a higher return if you sell part of a slice that is doing well (e.g. sell at the high) and invest it in one that is not (buy at the low) Because you do this based on some criteria, it helps circumvent the human tendency to 'hold on to a winner too long' (how many times have you heard someone say 'but it's doing so well, why do I want to sell now"? presuming trends will continue and they will 'lose out' on future gains, only to miss the peak and ride the thing down back into mediocrity.)

Depending on the volatility of the specific market, and the various slices, using re balancing can get you a pretty reasonable 'lift' above the market average, for relatively low risk. generally the more volatile the market, (such as say an emerging markets portfolio) the more opportunity for lift.

I looked into this myself a number of years back, the concensus I came was that the most effective method was to rebalance based on 'need' rather than time. Need is defined as one or more of the 'slices' in your portfolio being more than 8% above or below the average. So you use that as the trigger.

How you rebalance depends to some degree on if the portfolio is taxable or not. If in a tax deferred account, you can simply sell off whatever is above baseline and use it to buy up the stuff that is below. If you are subject to taxes and don't want to trigger any short term gains, then you may have to be more careful in terms of what you sell. Alternatively if you are adding funds to the portfolio, you can alter how your distribute the new money coming into the portfolio in order to bring up whatever is below the baseline (which takes a bit more time, but incurs no tax hit)

The other question is how will you slice a given market? by company size? by 'sectors' such as tech/finance/industrial/healthcare, by geographic regions?

  • HHH, That's why I suggested that within the results to focus on the academic stuff (easy to pick out based on the URL's that are .edu) and then dig within the references of those papers. Yes there's spam, always will be these days with all the SEO efforts. Even then, most of the 'spam' at least covers the basic theory, and some of them even include references to other articles – Chuck van der Linden Jul 1 '11 at 6:50
  • @Chuck van der Linden: If you mentioned the "inurl:edu" or "inurl:math" and warned about the spam, I would reconsider my downvote. Currently, it is hard to take this answer very seriously. – user1770 Jul 2 '11 at 18:25

Your Answer

 
discard

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.