Hi Guys,
Statistical models are a functional tool to help understand the interactions between complex social and physical phenomena.
The most common distribution deployed in this modeling is the Normal (Bell) curve. It’s a good choice for many phenomena, but has shortcomings when applied to investment annual returns, especially at the less likely outcomes that exceed the two standard deviation variation level.
I know, I know you are tired of me riding this hairy horse, but I promise this will be my last post on this matter for an extended time (but not forever). I was just jolted by a lightening bolt this morning, and wanted to share it with you. It could conceivably come to your rescue in your retirement planning.
Nassim Nicholas Taleb documented and named the impact of highly improbable investment events in his hugely successful 2007 book “The Black Swan”. That title alone captured the attention of a hungry public; it was sheer marketing brilliance.
He gave an electrifying name to events that were well known by scientists for many decades as the less exciting “Fat Tails”. Benoit Mandelbrot recognized these outliers in cotton market pricing, studied it for years, and published a superb book, “The (Mis)Behavior of Markets”, on the topic in 2004.
I attempted to incorporate Fat Tail elements when I generated a Monte Carlo code to explore portfolio survival prospects in the mid-1990s. I have long championed the advantages of using Monte Carlo-based analyses as an aid to the retirement decision task.
The scientific and engineering communities have been forever aware that not all physical events surrender to a Normal or Log-Normal statistical distribution. For example, I worked at GE for a short time, and within a week after my arrival, my section chief gave me a copy of a book titled “Statistical Models in Engineering”. I still have it. Various chapters are devoted to Normal, Log-Normal, Gamma, Beta, Rayleigh, Cauchy, Weibull, and other special statistical distributions.
Honestly, today I don’t know the merits, shortcomings, or applications of these numerous modeling options. These tools require specialized knowledge and considerable experience. That’s the bad news. The good news is that investors don’t need that mathematical level of sophistication. I even doubt if these distributions adequately capture real Black Swan events in a satisfactory manner.
My eureka moment was that I finally realized we can integrate Black Swan events into our retirement decision by experimentally using real world historical data in a random fashion. The really good news is that we can easily complete this task using a couple of options available on the Portfolio Visualizer Monte Carlo website that I recently recommended. Here again is the direct Link to that Monte Carlo simulator:
http://www.portfoliovisualizer.com/monte-carlo-simulationHere’s how to use the Portfolio Visualizer tool to estimate the impact of Black Swan events on your portfolio survival likelihoods.
Complete the short list of required inputs that reflect your holdings, goals, and time scale. In the Simulation Model box, Portfolio Visualizer offers three options: Historical Returns, Statistical Returns, and Parameterized Returns. For our current purposes, only the Historical Returns and Statistical Returns options need to be exercised.
The Statistical Returns do not specifically select Black Swan outliers, but only incorporate the smoothed interpretation of these data. The Historical Returns data set randomly selects from all historical data, so it includes the wild outliers specifically. The experiment is to run both distributions, and simply compare the outcomes. The impact of historical Monte Carlo events on your portfolio survival likelihood is the difference in the calculated probabilities.
I conducted a few experiments, certainly not comprehensive in scope.
Black Swans will lower the likelihood of portfolio survival by zero to only a few percent. Results will depend upon the specifics of your portfolio holdings, etc. I’ve run several test cases for a
50/
50 equity/bond mix that generated these sample outcomes. Retirement planning should include a sufficient safety margin to accommodate these surprises.
As a general observation, it appears that Black Swans are minor league players when the timeframe is long(I did most of my check cases with a 30 year time horizon). Sensitivity to Black Swans becomes more acute as timeframe is shortened.
Of course, this analysis only measures the impact of past Black Swans. No logical method can confidently project the frequency or magnitude of future Black Swans. That’s the nature of the uncertain investment beast.
I hope you visit the Portfolio Visualizer site to test the robustness of your portfolio and timeframe to Black Swan events. You just might learn something; I did.
By now, I’m sure you have tolerated enough of Monte Carlo and especially of me. So I’ll jump off the Monte Carlo bandwagon, at least for the moment. Thanks for your patience.
Best Regards.