You can download a PDF version of this commentary here.

Financial Advisors: In this piece, we run a risk-decomposition on a number of portfolio allocations.  If you’d like to have this risk decomposition performed on your portfolios, email us at with the subject “Risk Decomposition”. 


  • Last week’s commentary highlighted why we believe traditionally built portfolios may face return headwinds going forward.
  • Traditionally built stock/bond allocations also exhibit extremely high risk concentrations.
  • Non-traditional exposures, now available as low-cost ETFs, can help introduce non-standard risk exposures.
  • By combining non-traditional exposures, we can build a portfolio that has a similar overall risk profile as a traditional stock/bond portfolio, but with greater diversification in contributing risks.

In our commentary last week, we released a white paper discussing why we believe a traditionally allocated stock/bond portfolio may face headwinds over the next decade.

Most of our analysis focused on the return side of the equation, specifically discussing the impacts of low interest rates and slower GDP growth rates.

This week, we want to discuss the risk side of the equation.  Specifically, the high concentration of risk in traditionally allocated portfolios.

The Method of Risk Decomposition

To create a portfolio risk decomposition, we take the index returns (excess of the risk-free rate) of our target portfolio and regress them on a number of common risk factors, including:

  • Global equity returns
  • Yield curve level changes
  • Yield curve slope changes
  • Credit spread changes
  • Dollar index returns

This regression tells us the beta that our index has to each of these factors, as well as any return that is unexplained by them.

An important step in this process is culling non-significant factors.  In our process, if a factor’s significance level is greater than 1%, the factor is removed from the analysis and the regression is re-run.  This allows us to help reduce spurious variables in the analysis.

After performing this regression, we calculate an Analysis of Variance (“ANOVA”) Table (in this case, specifically a Type 2 table).

This table allows us to calculate the proportion of portfolio variance explained by each factor.

There are two important considerations here:

  • We use index returns of the portfolio to make sure we are looking at portfolio risk after the effects of diversification are embedded.
  • The results we are analyzing tells us information about portfolio risk breakdown, but tell us nothing about portfolio risk itself. Consider an extreme case, where we have a portfolio that is 99% cash and 1% equity.  While most would consider this a very safe portfolio, 100% of the portfolio’s risk will be explained by equity.  Why?  Because the cash has no volatility, so the volatility that is in the portfolio comes entirely from equity volatility.

Risk Decomposition for Traditional Asset Allocation Models

To examine traditional asset allocation models, we run this process on several asset allocation ETFs.  Specifically, we use the iShares Core Conservative Allocation ETF (AOK), the iShares Core Moderate Allocation ETF (AOM), and the iShares Core Aggressive Allocation ETF (AOA).

The proportional breakdown of risk for each portfolio follows:


Source: CSI Data and Federal Reserve Bank of St. Louis.  Calculations by Newfound Research.



Source: CSI Data and Federal Reserve Bank of St. Louis.  Calculations by Newfound Research.



Source: CSI Data and Federal Reserve Bank of St. Louis.  Calculations by Newfound Research.


The surprise here is that in a traditionally built portfolio, the vast majority of portfolio variance comes from the equity side of the equation.  In other words, even if you build a 20/80 portfolio, equity risk still dominates the variance the portfolio experiences.

This is a problem largely because it means the portfolio is highly susceptible to regimes where equity volatility spikes.

Is there another option?  We think so.


Risk Decomposition for a Non-Traditional Exposures

For investors willing to look beyond traditional asset classes, we believe there are a number of exposures that can be utilized to help balance the risk equation.

Consider, for example, the asset classes we utilize in our Multi-Asset Income portfolio.



Source: CSI Data and Federal Reserve Bank of St. Louis.  Calculations by Newfound Research.


As with any quantitative model, we think it is important that the numbers jive with our intuition.  We can see that asset classes like U.S. Dividend Stocks are dominated by equity risk, while U.S. Treasuries are dominated by rate risk.

Emerging Market Local Currency debt, on the other hand, introduces a significant amount of currency risk.  Yet U.S. dollar denominated Emerging Market debt does not.

These results align with our expectations and can highlight how different exposures can help introduce specific risk-factors into our allocation.


In our Multi-Asset Income portfolio, the strategic weights (i.e. the weights in the portfolio before we apply any tactical tilts) are governed by a yield-based Sharpe Parity process (like non-levered risk parity, but we incorporate expected returns using forward yield expectations).

The 12/31/2016 strategic model weights were:


Note: These weights do not reflect the Multi-Asset Income portfolios actual weights on 12/31/2016, but rather the strategic asset allocation the model has prior to any systematic tactical tilts employed in the portfolio process..

If we backtest these portfolio weights, we find that the historical volatility of the portfolio is slightly higher than the iShares Core Moderate Allocation ETF (6.81% versus 5.34%).

Yet despite this higher volatility, the underlying risks are noticeably more diversified:


Source: CSI Data and Federal Reserve Bank of St. Louis.  Calculations by Newfound Research.


So while the overall portfolio risk level is similar to the moderate portfolio, the sources of are completely different.

We believe that a more diversified risk decomposition means that the portfolio has a better chance of remaining stable across economic regimes.

The downside to this approach is that by diversifying risk more evenly across a number of exposures, it means the portfolio will have a higher degree of tracking error towards traditional asset classes.  So for investors that are highly sensitive to traditional reference points (e.g. the S&P 500), such an approach may cause undue anxiety.  In this case, we think a blended approach is best.



Traditionally built stock/bond portfolios may face a number of headwinds over the next decade.  From a risk management perspective, the biggest headwind may be the large equity risk concentration these portfolios have embedded.

Today, however, there are a large number of non-traditional exposures available in low-cost ETF wrappers that can help us more evenly distribute the sources of risk in our portfolio.

While this approach creates greater tracking error to traditional benchmarks, we believe this higher tracking error is simply the price we pay for better diversification and should be celebrated.  After all, low tracking error to traditional benchmarks implies high equity risk concentration.


Client Talking Points

  • Portfolio risk decomposition can help us discover which exposures are contributing the most to portfolio volatility.
  • Even very conservatively allocated traditional portfolios are dominated by equity risk.
  • By introducing non-traditional exposures, we can help balance out the sources of risk in our portfolio.


Corey is co-founder and Chief Investment Officer of Newfound Research, a quantitative asset manager offering a suite of separately managed accounts and mutual funds. At Newfound, Corey is responsible for portfolio management, investment research, strategy development, and communication of the firm's views to clients. Prior to offering asset management services, Newfound licensed research from the quantitative investment models developed by Corey. At peak, this research helped steer the tactical allocation decisions for upwards of $10bn. Corey holds a Master of Science in Computational Finance from Carnegie Mellon University and a Bachelor of Science in Computer Science, cum laude, from Cornell University. You can connect with Corey on LinkedIn or Twitter.