Author Topic: Does anyone actually use portfolio theory to determine asset allocations?  (Read 3070 times)


  • Pencil Stache
  • ****
  • Posts: 535
I’ve read some about it, and the concept is helpful for me. I get how you can reduce the variance of returns for a given expected return by selecting asset classes with low correlations. I’ve even found posted correlation coefficients, expected returns, and variances for various asset classes and plugged in the numbers for a few scenarios, just to try to understand it better.

But, it seems like all of the values you use for each asset class are estimated and uncertain, so any comparison of various asset allocations will also be very approximate. For example, even trying to use it to optimize a simple S&P500/bond allocation to within 5 or 10% based on your personal desired returns and comfort with volatility seems kind of ridiculous.

My personal approach has been to pick an asset allocation that seems “about right” and then not think too much about it. I’m sure there are calculators available, or someone could write a script to analyze 100’s of scenarios, but it seems to be more of an academic exercise than anything practical.

So, has anyone actually calculated expected returns and variance for their portfolio and used it to set their asset allocations?

Mr. Boh

  • Stubble
  • **
  • Posts: 149
  • Location: The Land of Pleasant Living
I don't use modern portfolio theory to determine asset allocation but I have looked into it. I think it helped me understand the idea of how an asset is correlated to other assets. I think that in turn helped me to better understand risk. In the end though, to me it just bolsters the old adage that diversification is the closest thing the investment world has to a free lunch.


  • Handlebar Stache
  • *****
  • Posts: 1463
... it seems to be more of an academic exercise than anything practical.

This part!

The basics: asset classes don't move the same way at the same time. Getting two or more assets that are uncorrelated or negatively correlated can help build a more efficient portfolio. In academic theory this is great and you could do a lot with it. However, in the real world, this is most relevant for diversification and then stocks VS bonds. High quality (traditionally government) bonds tend to go up when stocks go down and you can use that relationship to control the aggressiveness of your portfolio, and for obvious reasons you should be well diversified.

If you wanted to take that a step further, like the relationship between domestic and international stocks, you could do a lot of research or you could just read whitepapers from major investment firms like Vanguard. Their analysts already did the research so you don't have to.


  • Handlebar Stache
  • *****
  • Posts: 2292
  • One Does Not Simply Work Into Mordor
Yup. [when I said yup I meant agreement with the replies above] Bernstein covers this thoroughly in "The Intelligent Asset Allocator." This method tends to strongly overweight whatever did best in the most recent period, which then more often than not goes on to do very poorly over the next period. Whenever a trend has become strong enough to be actionable, it will suddenly stop paying off. Eventually you are correct that all you can do is use some general principles and then hang on.

And this is where you can play around with it.
« Last Edit: January 01, 2018, 09:48:20 PM by Radagast »


  • Handlebar Stache
  • *****
  • Posts: 1941
I definitely use portfolio theory to determine asset allocations but I don't think I view portfolio theory the same way as the op.

Modern portfolio theory to me just says a diversified portfolio across different asset classes and within those asset classes works. I've read a lot of Bernstein's books and recently I read McClung's book. The ideas within those books have guided how how I've constructed my portfolio.

I definitely don't try and optimise my asset allocations. I don't think that works.


  • Bristles
  • ***
  • Posts: 422
Add more bonds for less volatility. It's that simple.


  • Pencil Stache
  • ****
  • Posts: 535
Thanks for the book recommendations! I may check them out some time, if I get curious. From all the replies so far, it doesn’t sound like anyone else spends much time trying to fine-tune asset allocation down to the last percent either.


  • Walrus Stache
  • *******
  • Posts: 5830
One problem with modern portfolio theory (MPT) is how easily the suggestions vary depending on the start, end, and asset classes included.  Running a calculation from 1970-2010 will probably be different than the past 25 years (1993-2017), especially if you split stock allocation into: US large, US small, international developed stock, and emerging markets. 

I think there's plenty of evidence that U.S. and international stock markets have become more correlated over time (as international businesses become more common and intertwined).  With MPT, everything relies on the correlations.  When correlations change over time, allocations change along with them.  Which means data from long ago, when U.S. and international were less correlated, may not be relevant today.  And could also explain why MPT provides varying answers based on start and end dates (when you have several stock asset classes, like emerging markets and US small cap).


  • Handlebar Stache
  • *****
  • Posts: 1195
"Modern Portfolio Theory" is a broad term involving several different mathematical risk management ideas in investing.  I personally think that efficient frontiers are not always actionable and that fixating on expected returns is unnecessary if you prioritize consistency.  But the fundamental insight that one can effectively manage portfolio risk via intelligent asset allocation is spot on and I use that concept all the time to study portfolio construction ideas.  IMHO, the trick is to think more in terms of managing uncertainty rather than maximizing returns with any predictive precision.


  • Handlebar Stache
  • *****
  • Posts: 2151
  • Age: 63
  • Location: Redmond, WA
    • Evergreen Small Business
I try to employ logic of MPT.

For what it's worth, I like the Bogleheads philosophy and used it long before Bogleheads even existed... but one of the flaws with the strategy is the undercurrent of anti-MPT thinking you regularly see:

FYI, Vanguard has a good whitepaper on the issue of dynamic correlations here:

Also Tyler has done a very nice job talking about how portfolio variability impacts your SWR at link below, which is another thing people often (IMHO) don't think enough about:

P.S. Bernstein's book, "Rational Investor" does a great job stepping you through the numbers and then also talking in practical terms about why MPT works better in theory than in practice:


  • 5 O'Clock Shadow
  • *
  • Posts: 1
Re: Does anyone actually use portfolio theory to determine asset allocations?
« Reply #10 on: January 03, 2018, 01:38:08 PM »
There is an app called Stock Optimizer that uses MPT to calculate optimal portfolios of stocks in S&P500, NASDAQ and Dow Jones. Don't know if they use is for asset allocation just yet.

Link below:


  • Magnum Stache
  • ******
  • Posts: 2995
  • Age: 40
  • Location: North Carolina
    • Freedom35
Re: Does anyone actually use portfolio theory to determine asset allocations?
« Reply #11 on: January 03, 2018, 10:49:30 PM »
Do you include the Fama-French model in MPT?


  • Pencil Stache
  • ****
  • Posts: 535
Re: Does anyone actually use portfolio theory to determine asset allocations?
« Reply #12 on: January 03, 2018, 11:38:03 PM »
I guess this could be a more general question about using some sort of mathematical approach to try and calculate an “optimal” asset allocation for a portfolio, as opposed to using general recommendations or rules of thumb.

To answer your question: I have not really looked deeply at the Fama-French model.

« Last Edit: January 03, 2018, 11:48:24 PM by Wintergreen78 »


Wow, a phone plan for fifteen bucks!