Effective August 18, we have adjusted the prices at which we assign the Morningstar Ratings for stocks on the equities that we cover. This move is based on our uncertainty about the valuation of the stock, as well as some changes to the price thresholds at which we first identify stocks as overvalued or undervalued. We think of the new rating cutoffs as fine-tuning of our existing system to best reflect a balance between theory and empirical results, rather than a shift in the rating methodology.
For any given company, we consider our estimate of its fair, or intrinsic, value to be a single point within a range of possible fair values. To help us get a handle on that range, we perform scenario analysis on individual companies, stress-testing our valuation models under a variety of assumptions. We then assign that company an uncertainty rating based on this assessment. For low-uncertainty stocks, we believe we can bound the values of the company to a relatively tight range. Therefore, we demand a relatively small margin of safety before we feel confident that the shares are undervalued, which would lead us to recommend purchase of the shares. For high-uncertainty stocks, on the other hand, we can only assign a wide band around the values of the stock, so we demand a steep discount to our fair value estimate before we are confident that the shares are undervalued enough to deserve a 5-star rating.
During the depths of the financial crisis in 2008, we widened the bands to reflect the heightened market volatility. By increasing the margin of safety, we were able to increase the quality of companies landing on our 5-star list. Even so, we had more than 600 stocks rated 5 stars during the market meltdown--a level we thought appropriate, given how cheap stocks had become. Then in late 2010, as market conditions normalized, we adjusted our overall high- and very high-uncertainty bands to bring them more into line with the other bands.
Now, after further examination of the theory and performance of our system, we are making some adjustments to the prices at which we assign our star ratings. We've made two types of changes: We've lowered the premium to our fair value estimate that we require before we declare that high- and very high-uncertainty stocks should be sold, and we've narrowed the premium and discount to our fair value estimate before we first flag stocks as overvalued or undervalued. The changes are highlighted in the below chart.
We changed our 1-star rating threshold to better reflect the empirical data regarding the distribution of stock prices three years in the future. We had previously set the thresholds for saying that a stock was overvalued and should be sold based on the theory behind option prices, namely that outcomes for stock prices are log-normally distributed. In the empirical data we gathered regarding the distribution of stock prices for stocks grouped by our uncertainty rating, shown below, the distribution of outcomes fall approximately midway between log-normally distributed and the normal Gaussian distribution. For example, when we grouped low-uncertainty stocks and looked at the distribution of prices three years in the future, the bottom 5th percentile stock traded for 40% of the price of the median stock at the end of the three-year period. Our data set starts with our initiation of the business risk rating in 2001 and ends in 2007, with subsequent three-year price distributions ending in 2010.
With this change, we formally set the price threshold before we assign a 1-star rating to the 75th percentile of a distribution that is halfway between a normal distribution and a log-normal distribution, rounded to the nearest 5%. Because of the small differences between normal and log-normal distributions for low- and medium-uncertainty stocks, the small changes in our methodology are hidden by rounding the results to the nearest 5 percentage points.
We changed our 2- and 4-star rating thresholds for both qualitative and quantitative reasons. Qualitatively, the narrower bands will allow us to highlight more potential investment opportunities when markets as a whole are more fairly valued. Quantitatively, we tested a trading strategy of buying stocks when we assign a 4-star rating and selling a stock when we assign a 2-star rating, and the strategy was more profitable with the narrower thresholds relative to fair value than with the thresholds being replaced.
The result of the most recent change will be a somewhat lower hurdle for stocks of companies in the high- and very high-uncertainty bands to hit 1 star. Practically speaking, this change will mean slightly more 1-star stocks than we would have had otherwise. Also, the hurdle for 2- and 4-star ratings will be lower, and the practical result is likely that more stocks will be identified as 2 or 4 stars, representing slightly overvalued and slightly undervalued positions, respectively.
The new thresholds at which we assign the star ratings based on price/fair value and uncertainty rating are shown in the chart below.
