Static CMAs models assume that markets follow a “random walk.” Under random walk assumptions, past returns have no impact on future returns. A long series of positive returns does not increase the probability of a negative return, just as a long series of “heads” when flipping a coin does not impact the results of the next coin toss. Asset class returns, like the results of a coin toss, are assumed to be random.
The probability of a coin landing on heads is always 50/50. Similarly, under random walk assumptions, the best estimate of future returns is assumed to be the long-term historical average (e.g. return assumptions are “static” near average return levels). Returns for a given period may be significantly higher or lower than historical averages based upon the riskiness of the asset class, as measured by its volatility. These fluctuations around the expected return are assumed to be driven by the random variations of a particular time period, just like when flipping a coin you might get 10 heads in a row purely by random chance. However, even after 10 heads in a row the odds of another heads remain 50/50. Similarly, no matter how high returns have climbed during a long bull market or how low they fall during a market crisis like 2008, Static CMA models assume that expected returns remain close to the long term historical average.
The chart below shows the historical support for the random return assumption underlying Static CMAs. The chart plots 1-year large cap “real” return (returns after accounting for inflation) against a simple measure of valuation – distance from trend (DFT). DFT first calculates the long term trend return for an asset class, and then measures how far a long series of positive returns (heads) or negative returns (tails) have driven valuation above or below that long term trend. As shown on the chart, whether markets start well below (undervalued) or well above (overvalued) the long term trend, 1-year returns appear to be a random scattering of outcomes with most results ranging between about -40% and +40%. Only at the very extremes of -50% or +100% DFT does valuation appear to impact returns, and there are very few data points at those extremes (it could be random noise). In other words, short term returns make a credible argument for random walk assumptions.