Co-founder of @AuroraSignals, quant hedge fund co-manager, financial advisor, entrepreneur, trader of equities | forex | options

Tue 04 November 2014


Twitter Logo Facebook Logo Google Plus Logo LinkedIn Logo Mail Logo

"Diversification": Why your portfolio probably doesn't have it

In a world of ever-present risks and uncertainties, effective diversification isn't just cool to have, it is essential.


Source:Randy Clasbergen

Diversify. Diversify. Diversify.”

We’re sure you’ve heard the saying before. Many investment advisors and even market experts tout the importance of diversification when it comes to investment portfolios. Unfortunately, the industry standard to properly addressing diversification is often insufficient. How do most asset managers attain diversification for their clients’ portfolios? Common practice is to look at correlation between assets within a portfolio. What does this mean? Put simply, correlation is the relationship between two or more assets over a given time period. Correlation aims to answer the question, “when Stock A rises 2%, what does Stock B do on average?” While this is an oversimplification, it’s not far from the truth. To get an answer, asset managers use historical relationships among two or more assets (stocks, bonds, etc.). The problem: this creates an assumption that the relationship (correlation) among the assets will continue into the future. What this really means is asset managers are assuming historical correlation will hold going forward, which takes a bet that future economic conditions will look like the past. This is rarely true, which is why correlations are fairly unpredictable—leading to hidden risks being taken on by the advisor.

Does correlation exist?

Ray Dalio, manager of $120 billion hedge fund Bridgewater Associates, explains that most people, even experts, approach the problem incorrectly.

"People think that a thing called correlation exists. That’s wrong…Correlation is just the word people use to take an average of how two prices have behaved together. When I am setting up my trading bets, I am not looking at correlation; I am looking at whether the drivers are different.”

-Ray Dalio, Hedge Fund Market Wizards

The issue with many industry standard approaches is advisors often claim they’re providing diversification by investing their clients’ in many different stocks across numerous sectors. Simply placing client funds in a basket of 25+ stocks does not provide proper diversification—especially if the strategy is based off buy-and-hold tactics or is a non-hedged portfolio. US stocks historically have had an average correlation to one another of 60-80% (see Figure 1). During market crashes the correlation among stocks can rise to nearly 100%, meaning most stocks move in lockstep during a crash. These correlations are so high that such a portfolio is simply not diversified in a manner that helps mitigate risk--which is the entire point of diversification in the first place. “Diversification” fails to protect portfolios when it is needed most. This is a serious problem and many investors (and their advisors!) are simply unaware of it.


Figure 1: Rolling correlation of S&P500 and a "diversified portfolio". The diversified portfolio in this example was constructed using a traditional buy-and-hold 70/30 stock/bond allocation. 70% of the portfolio was comprised of equally weighted allocations of the 10 major stock ETF sectors as well as gold and silver ETFs (GLD, SLV). 30% of the portfolio was comprised of equally weight allocations of the top 3 bond ETFs (BND, AGG, LQD).

Does a diversified buy-and-hold portfolio work?

Even experts admit standard portfolio diversification approaches often fail to protect during market crashes. Don't believe it? See what Yale and Harvard have to say about their experiences in the 2008 crash with their sophisticated multi-billion dollar endowment funds:

“During this period [2008—09], equity exposure hurt results, diversification failed to protect asset values and illiquidity further detracted from performance.” 1

-David F. Swensen, Yale Chief Investment Officer

“With a few notable exceptions, nearly every asset class did poorly … Our real estate portfolio, for example, suffered a loss of over 50% during the year … While diversification has been a mainstay and a driver of the portfolio’s return over the long term, the benefits of diversification did not bear out through the rapidly evolving and widespread events that unfolded in FY 2009.” 2

-Harvard University

Why is it that even some of the brightest minds in the industry get duped by market crashes when they've been properly following an investment approach that has been around for decades? Because it only works when markets behave normally, and fails miserably when they do not.

The most common investment models target optimized allocations of stocks, bonds, and cash. Remember our article on risk? Advisors typically produce a pie chart showing allocations of stocks and bonds (i.e. 80% stocks, 20% bonds) based on a client’s risk tolerances (Figure 2). Why? Because it’s the industry standard, and it’s very simple. Some advisors claim this to be a “sophisticated investment” approach, but in truth is anything but sophisticated. Does this approach really mitigate risks or protect portfolios in a market crash? Almost never.

Pie chart

Figure 2: Traditional pie chart of stocks, bonds, cash allocations. Typically, the diversification breakdown is by sector, company, bond duration, credit rating, etc.

Garbage in garbage out

Before an advisor can perform a computer optimization for asset allocation (i.e. which stocks or bonds to buy more or less of), they must first select the group of assets to analyze. The issue with this is if there is no true systematic approach to selecting the assets then the next step, allocation optimization, can become somewhat irrelevant. It is a simple problem of “garbage in garbage out". If an advisor or individual chooses a poor group of stocks and bonds, then regardless of how they are allocated, the portfolio will most likely underperform to expectations. The value from taking a more systematic approach comes from the removal of arbitrary/emotional decision-making. The way assets were selected one quarter will be consistent in the following quarter. Unfortunately, this is only part of the problem. Even with effective asset selection, the issue of diversification is not always properly addressed. Below we explain why.

How is a typical portfolio allocation constructed?

  1. In more recent times (thanks to technology), the advisor uses a computer program that looks at the historical correlations and returns of this group of stocks and bonds.
  2. The program then produces a chart displaying “ideal” risk to reward allocations of stocks and bonds based on the client’s risk tolerances and desired annual portfolio return (see Figure 3).

Efficient frontier

Figure 3: The Efficient Frontier was designed help asset managers construct a set of optimal portfolios that offers the highest expected return for a defined level of risk or the lowest risk for a given level of expected return. Empirical research shows this model takes on many dangerous assumptions that do not properly address the issue of diversification, however.

The allocation problem

To produce these “ideal” diversified allocations, the system relies on historical correlations and returns and takes the assumption that this holistic snapshot will continue into the future. This approach places heavy reliance on these assumptions and solves for just one problem: allocation. There is little evidence the issue of risk mitigation is properly addressed, as they claim. This is where it becomes dangerous, believing you have addressed most risks when in fact you have not—this actually adds hidden risks--great, so now the problem is worse than when you began! Taking a few historical examples of portfolios constructed using the above method shows that in events like the 2008 crash, this approach failed to mitigate risk and caused portfolios to sustain massive losses that took over 3 years to recover from (see Figure 4). So we ask, how well did “diversification” really work in that case?

Industry standard

Figure 4: Comparison of S&P500 and industry standard buy-and-hold 80/20 stock/bond portfolio approach using Mean Variance Optimization for the individual ETF allocations (MUB, LQD, PCY, VTI, VTV, VOE, VBR, VEA, VWO). If you were unlucky enough to of bought in 2006 or 2007, by the lows of 2009 you're looking at portfolio drawdowns of 40-60%! Note: a 50% drawdown will require you to make 100% just to get back to even--which is the exact case for many unfortunate investors during this crash.

"Uncertainty" isn't in Diversification's vocabulary:

Even when advisors diversify a portfolio based on sector, company, or even geographical location, it doesn't fully address the issue of downside risk mitigation--or even worse, market uncertainty (aka: unknown unknowns). Having all of your money in one stock, i.e. Exxon, is downright dangerous (I've actually seen this before). Having it spread among 5 is safer, or even 40, but it still doesn't protect you from systemic risk (total market crash). Add in some fixed income and you're getting better; alternative assets such as commodities or private equity help even further--but this starts to get outside the scope of what most advisors do for their clients...especially client's with small portfolios. Even then, you still have the problem of uncertainty (the unfathomable), which is the case of a global market meltdown--similar to what we saw in 2008--where nearly every asset fell. None of these solutions completely address behavioral risk, which is quite impossible to predict, or model. So what do you do, give up? Not quite. What if I told you the mere fact of understanding all of the above already helps you address part of the problem. False confidence is one of the greatest dangers to anyone's investment portfolio and it's why many professionals are actually amateurs when it comes to dealing with this issue.

Industry standard


"If you think it's expensive to hire a professional to do the job, wait until you hire an amateur".

-Red Adair

Address the problem from a different angle

While no approach is perfect, sometimes the best approach is to look at what not to do (via negativa), and then go from there. Having knowledge enough to avoid making stupid decisions or taking false assumptions can have a greater outcome than one would expect. When it comes to our investment approach, we don’t construct portfolios based on stock and bond diversification. Why? Our method for portfolio management does not involve optimizing a portfolio merely in the holistic “diversification” sense. The problem to address is not just a portfolio-level problem. The problem to address is more of a problem on an individual-asset level. What does this mean? Remember Ray Dalio's quote above? In just one sentence, he explains a very direct way to go about this problem: look at the drivers, not the correlation. Drivers are more consistent (and sometimes 100% concrete) in comparison to correlations. Once you have selected your assests through individual analysis, you can then zoom out a step and analyze at the portfolio level--creating a macro view of what is going on.

Individual asset analysis.

Separate the trees from the forest. Where the traditional approach constructs a model for a given portfolio, we aim to construct a model (or strategy) for every single asset we hold in our portfolios. Not only does this approach allow one to take less assumptions (i.e. projections of future asset correlations and returns), it also allows one to analyze and manage every single asset in the portfolio--regardless of what the others around them are doing. This systematic approach helps better address the “garbage in, garbage out” problem by individually and systematically assessing every asset chosen for the portfolio. Correlation and expected returns of the holdings are no longer factors in ones decisions because the emphasis is placed on causation; or simply put, the cause and effect of certain factors on each individual position--that is how you get a better picture of what's actually going on. Apply this method to various assets and income streams and you've now created a much more robust portfolio in comparison to most others. To take it a step further, going beyond just addressing risk, the issue of uncertainty (the real portfolio killer) must be addressed with the use of portfolio hedging. You can have a massive basket of various assets, but if you aren't hedging the downside, systemic crashes can do serious harm to your portfolio.

The takeaway points

  1. Correlation is a misleading metric for understanding portfolio risk.
  2. Traditional buy-and-hold strategies get pummeled in crashes.
  3. Long-only optimal asset allocation strategies also get pummeled.
  4. Garbage in garbage out: choosing the proper assets is a crucial to the problem.
  5. Uncertainty is often ignored in the world of "diversified portfolios."
  6. Look at the drivers of your assets, and analyze them individually.
  7. Hedge your portfolio if possible, which helps protect against unknown risks.


  1. Yale Endowment Report, 2009
  2. Harvard Management Co. Endowment Report, August 2009.
Go Top
comments powered by Disqus