Monte Carlo Simulations are Not Perfect
Monte Carlo simulation is a complex statistical modeling method which can be useful in financial planning. The goal of Monte Carlo analysis is to apply statistical realities over multiple simulations and conclude how many of those iterations had a successful result. As I have said many times, financial planning is as much an art as is it a science. Unfortunately there are no absolutes or guarantees even with proper planning. Past performance is not an indicator of future results, so relying on historical averages can be misleading. Analysis is a tool, but only the proper use of the tool combined with skilled interpretation provides any value.
The May 2nd article, "Odds-On Imperfection: Monte Carlo Simulation" in the Wall Street Journal explores using Monte Carlo analysis as a planning tool. Specifically, it points out that a dramatic downturn in the market like the one we have been living for the past year or so, would generally be ignored in a Monte Carlo simulation since it is statistically unlikely to occur. The article argues that once-in-a-lifetime market events are occurring with greater frequency and therefore the best analysis tools should give catastrophic episodes greater importance.
One solution to the problem is to continue using Monte Carlo analysis with "fat-tailed" distributions. Ordinarily, extreme events occur on the outer edges of a bell shaped curve. The more extreme, the less likely it is to occur. A fat-tailed curve allows for a greater probability of an extreme downtown than an extreme period of growth. Another solution is to use a large number of simulations in a Monte Carlo analysis. The article suggests using 10,000 simulations or more.
Some critics think these limitations of Monte Carlo analysis prove that it is an invalid tool. I disagree. Monte Carlo analysis has never been perfect, and it never will be. As we grow in our experiences it is only natural that we discover limitations that we had previously not considered. The wisdom is in learning and adapting to make the tool better.
As for my clients, the Monte Carlo simulations I use already employ 10,000 iterations. According to the manufacturer of the software I use, they will be making "fat-tailed" distributions available in a few weeks. This has the potential to make the tool better, but still imperfect. No one can predict the future financial markets with certainty, but we can use the best tools available to plan for potential scenarios.