How To Pick The Best Asset Allocation Model

by: Tony Ash


Given the preponderance of "big data," investment PhD's, and massive computing power, you would think there would be one "best" asset allocation model for a given level of risk.

There isn't!

It is best to look at various analytical tests of a strategy and use the output judiciously to help you make your decision.

Given the preponderance of “big data,” investment PhD’s, and the massive computing power that we have today, you would think it would be easy to come up with the “single” best asset allocation for a give risk tolerance. However, from my previous article here, the facts show a large dispersion of returns from the 700 funds that play in the targeted-risk asset allocation space per the Morningstar 50%-70% Equity Category. As a Part 2 follow-up to that article, following are some of the tools that I use to help evaluate the robustness of a risk-targeted asset allocation.

The idea here is to identify a risk-targeted long-term strategic asset allocation to get “core” market exposure. A typical investor will have many unique considerations that will dictate the need for personalized “satellite” exposures in addition to a core portfolio. These personalized factors could include asset allocations to support needs for guaranteed cash flows, high liquidity or low drawdown risk and this is outside the scope of this article.

Firstly, we need to recognize the overall poor performance of “active” managers over the last business cycle. According to S&P Dow Jones Indices, over a five-year period ended 2016, 88% of large-cap, 89% of mid-cap, and 97% of small-cap managers have underperformed their respective benchmarks. We need to immediately discount the vast majority of active “stock pickers” and focus more on “passive” index-based strategies to reduce the risk of underperformance from active managers.

Not to be overlooked in the active manager universe, of course, are the much higher expense ratios of active managers compared to index-based strategies in the ETF form. Per the ICI, the simple average equity mutual fund expense ratio was 1.28% in 2016 compared to about 0.10% (or lower) for typical index-based ETFs from Vanguard, Schwab, iShares, and Fidelity. An approach to aid in success is to avoid active managers due to the historical underperformance and their much higher expense ratios.

Once we focus on index-based assets and exclude the active “stock picker” bias, we can now use some analytics to help in the decision process. A good rule of thumb is to use the classic “60/40” equity/bond asset mix as a baseline upon which to compare alternative strategies.

It is relatively straight-forward to find a vendor to provide a “moderate” target risk asset allocation. One can easily visit any of the robo-advisers (Wealthfront or Betterment) or ETF sponsors (Vanguard or iShares) to see what they recommend for free. You will note that the asset allocations all trend to different approaches with enough material differences to be significant. For example, Wealthfront is an advocate of tax-free municipal bonds for taxable accounts due to their current advantageous pricing, whereas Vanguard and iShares ignore that space. Likewise, some providers take a more detailed approach with more focused ETFs compared to a more general approach with fewer more general ETFs.

I used the iShares asset allocation generator here. It is easy to use and generates a good simple mix of ETFs for a basic core strategy. Of course, since it is sponsored by iShares, the recommended ETFs are all under the iShares label. However, it is a simple exercise to isolate out the target asset classes and then choose your own favorite index-based passive ETF from Vanguard, Schwab, etc.

The Moderate risk asset allocation from iShares follows in the table below. It ticks off all the typical equity and bond asset classes that would be included in a moderate risk portfolio, but excludes other potential asset classes like high yield bonds, preferred stocks, inflation bonds, municipal bonds, or gold. It is still amazing to me that you can get such broad market exposure at such a low fee of only 7 basis points!



Exp. Ratio


iShares Core S&P 500 ETF




iShares Core S&P MidCap ETF




iShares Core S&P Small-Cap ETF




iShares Core U.S. REIT ETF




iShares Core MSCI EAFE ETF




iShares Core MSCI Emerging Markets ETF




iShares Core Short-Term U.S. Bond ETF




iShares Core 5-10 Year USD Bond ETF




iShares Core International Aggregate Bond ETF




Avg. Exp. Ratio/Total



So, how comfortable are we with this moderate risk portfolio? How would it have performed over the past years with backtesting? How could it perform going forward? To answer these and other questions, I used, a free website that does some pretty neat analytics. It offers a broad menu of potential analytic tests, but I focused on only three things for simplicity: a 15-year backtest, a forward-looking Monte Carlo simulation, and a simple Markowitz Model efficient frontier analysis.

Because some of the iShares ETFs have relatively short inception dates, I chose to model the broad asset classes instead of the actual ETFs. Since the expense ratios are so low (only 0.07% for the portfolio), there would not be much difference as it relates to interpreting the results.

The table below highlights the key backtested investment statistics for the iShares Moderate portfolio compared to the 60/40 portfolio over the past 45 years starting in 1972. It includes annual rebalancing to targets, an especially important feature given the long time horizon, and no fee structure. As can be seen, the iShares portfolio underperforms the 60/40 portfolio by an annualized 0.72%. The iShares portfolio also has a higher standard deviation, max drawdown, and lower Sharpe ratio (as a measure of excess risk per unit of return). All in all, seems like a weak strategy compared to the simplistic 60/40 based on these metrics.



Std. Dev.

Max. Drawdown

Sharpe Ratio

iShares Moderate





60/40 Baseline





The next test I used was Monte Carlo simulation to see how a $1,000,000 portfolio could perform in the future based on historical returns. I used a 15-year projection period with annual rebalancing with no contributions or withdrawals for 10,000 iterations.

The table below summarizes the results of the Monte Carlo simulation. Just like the backtesting results, the iShares portfolio underperforms the 60/40 portfolio for all of the measured percentiles. At the 50th percentile, the median performance, we can see that the iShares portfolio has a terminal value at the end of 15 years of $3.072 million compared to $3.585 million for the 60/40 portfolio.







iShares Ending Balance






iShares Max Drawdown






60/40 Ending Balance






60/40 Max Drawdown






Finally, based on the Modern Portfolio Theory and the Markowtiz efficient frontier model, let’s see where the iShares portfolio fits compared to an “unconstrained” portfolio. Based upon historical data from 1995 through 2017, the iShares portfolio would generate a modeled expected return of 8.12% with a 10.07% standard deviation compared to the unconstrained efficient frontier model with a higher 10.43% return for a similar level of risk. Consequently, the iShares portfolio falls below the efficient frontier. However, of course, due to the unconstrained nature of this model run, the efficient portfolio seems a bit “weird” in that it is comprised of only three asset classes: US Mid Cap, US Intermediate Treasuries, and US REIT. As is common with this modeling approach, however, we could constrain the model to get more realistic results, but then we would be using our personal biases to influence the results.

So, what does all this mean? Do we exclude the iShares Moderate portfolio from consideration or accept it despite all of its losing scores?

Certainly, the iShares Moderate portfolio looks like a very typical moderate risk portfolio. It has all the “right” broad asset classes and all the weightings seem about right. The one thing we need to be especially wary of is the “modeling” risk we are exposed to here. Though we all know “past performance is no guarantee of future results,” the Monte Carlo and efficient frontier models are using past performance as a guide to the future. Additionally, I chose generic time frames (i.e., 15 years for the Monte Carlo and 22 years for the efficient frontier) that could be impacting the output. In fact, tweaking the model runs to shorter/longer time horizons narrowed the gap of underperformance. The modeled output is interesting, but is by no means conclusive. This fits the category of “not being a slave” to your models and knowing how to use the output. It is likely that the iShares Moderate portfolio is within a performance tolerance that makes it indistinguishable from its naïve 60/40 baseline.

Given the weaknesses of the models and how some of the factors could be distorting the model results, I feel that the iShares Moderate risk model portfolio is within a range of outcomes that qualify it as a buy. As mentioned at the beginning, this portfolio could be used as a core with additional satellite exposures to customize to your unique personal situation.

Most interestingly from this review, the 60/40 portfolio appears to be a prime candidate for selection due to its great historical performance stats that the iShares Moderate portfolio had trouble beating. Most investors, however, would run away from the 60/40 portfolio due to its low level of asset class diversification and too strong US bias at this point in the time.

Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.