Target Practice Is A Good Way To Measure Portfolio Accuracy


I don’t think many people will be surprised to learn that I’m not a fan of using single averages to describe portfolio returns or over-simplified metrics like standard deviation to measure risk.  Both of those numbers obscure so much information that they often lead to disappointing investing decisions and give backtesting a bad name.  I personally have a much more nuanced perspective that embraces the idea of unpredictability in financial planning while keeping things in terms that make sense in the real world, although I’m well aware that uncertainty is a very difficult concept to understand and an even more difficult one to apply.

So sometimes even the simplest questions can trip me up.  For example:


What expected return would you use for the Three-Fund Portfolio for your own financial planning?


One way to answer that without attempting to predict the future is to point to the longest average return I have data for, but that has a notable drawback.  Just like how the most thrilling roller coaster in the US averages a speed of only 22 mph, single long-term averages completely lie about the intermediate experience of the ride.  And because of that I usually roll my eyes when people quote the very long-term average return of stocks or bonds since the late 1800s as if they know a single person who plans to invest for more than 100 years before putting their money to good use.  So expressing the uncertainty of returns in realistic investing timeframes for normal people is important to me.

Another way to answer is to point to tools like the Heat Map or Rolling Returns calculators that display the historical outcomes for every timeframe we have data for.  While they are excellent at painting a full picture for the ups and downs of a portfolio and can even be used to match investing goals to timeframes, I admit that they do not necessarily provide a straightforward answer.  Identifying an individual return using that information requires a bit of personal interpretation, and I’d love to be able to put a firm stake in the ground to establish appropriate expectations.

Striking that balance between desired specificity and realistic uncertainty is tricky to navigate, but it’s not like it has never been done before.  For a good example, let’s turn to archery.

Let’s say you are personally charged with selecting an archer to compete in a shooting competition.  The archer is your portfolio, and the target is your future financial goal.  While knowing full well that you will never be able to perfectly predict the final location of a specific arrow, you also inherently understand that Katniss Everdeen would be a better choice than a random guy off the street who has never seen a bow before.



So how do you identify the most accurate archer ahead of time?

You could look at the location of their last shot, but it’s possible the rookie just got incredibly lucky.  You could ask about their lifetime average location, although that says nothing about how often they missed the target entirely.  You could even just default to the most popular archer with the highest name recognition, although for all you know he simply paid a lot of money for good press.

Personally, I’d set up a target at a reasonable range and give each candidate a quiver full of arrows to show me what they’ve got.  The pattern of holes in each target would give me a realistic picture of the accuracy of each archer allowing me to select the best one for the job.

What happens if you apply the same idea to asset allocation?  You get the Target Accuracy calculator.




Think of the Target Accuracy calculator as a financial archery range with the target set 15 years out.  Each shot is a real-life historical result for an investor starting in every possible start year from 1970 to the present.  By studying the cluster of holes in the target, one can get an excellent idea for the realistic range of long-term returns for a particular portfolio.

While it’s built on the same foundation as the Portfolio Growth calculator, the Target Accuracy calculator also has a few additional tricks up its sleeve.

First, it automatically filters out the top 15% best returns and bottom 15% worst returns.  Every archer gets lucky or unlucky sometimes, and by focusing on the middle 70% of historical outcomes it provides a decent idea for a reasonably appropriate range of returns excluding the most extreme outliers.

Next, it does a little extra legwork and calculates the real CAGR for each end of that middle band. So the range of the most common annual returns for a portfolio is right at your fingertips with no other tools required.

And perhaps most importantly, you’ll notice there’s an additional asset allocation column available.  Change it to anything other than zero, and it allows you to directly compare the accuracy of two portfolios in a single image.  This was inspired by some charts I made a while back for a post on the benefits of diversification, and I find it really useful for explaining the relative returns and uncertainties of two portfolios.

For example, here’s a chart that overlays the Classic 60-40 and Total US Stock Market.




At first glance, it’s easy to see that the Classic 60-40 was more accurate than the Total Stock Market portfolio.  However, the price to pay for reducing volatility with bonds was lower returns on the high end while leaving the low end unaffected.

For another example, let’s look at the Golden Butterfly vs. the Three-Fund Portfolio.




Surprised?  Just because one portfolio had a narrower spread of returns than another does not automatically mean that it had a lower average return.  Intelligent asset allocation opens up all kinds of possibilities for productive risk management.

So circling back to the original question, what would I personally use as the expected return for the Three-Fund Portfolio for my own financial planning?  According to the Target Accuracy calculator above, a reasonable range of compound real returns excluding the most extreme outcomes was between 3.8% and 8.6% a year for a long-term investor.  That doesn’t mean the future return can’t be higher or lower than that, but like archers with a proven track record portfolios generally do not change their nature overnight and planning for something in that range is backed by good data.  And if it were me, I’d use the low end of the range for conservative market projections while adjusting my savings rate accordingly.  If that target goal is an especially important apple sitting on your own head, nobody will complain if the arrow sails high but a miss low could be a major problem.

No portfolio is perfectly predictable, but not every portfolio is equally unpredictable.  Take the time to study the target accuracy of your own portfolio and you’ll be one step closer to a dependable plan to achieve your important life goals.



On a quick administrative note, you may also notice that the Benchmark calculator is no longer listed on the site.  The Target Accuracy calculator is an upgraded version of the Benchmark calculator that displays every start year simultaneously instead of just one at a time.  That said, if you miss a feature of the Benchmark calculator or have suggestions for a new tool, please don’t hesitate to contact me.