Nerdy Stuff Matters to Making Money!
Before we show you how quality and reasonable valuations are the breeding grounds of greatness, let’s talk about a common analytical error.
When analyzing stocks and investment strategies, using bias-free data is a basic and critical first step. For a beautiful explanation, we guide you to Teddy Koker, a researcher at MIT’s Lincoln Labs, elegant walk-through, complete with programming instructions. For those interested in a less complex example, this post is for you.
A friend sent me the image below with the affectionate note: “Nerd, this popped up in Facebook, seems like something you would find funny.” She was, of course, correct. We will use a few bullets to go from a damage analysis of World War II planes to how survivor bias in finance creates havoc for investors.
Survivorship Bias Plane Explained – What is Survivorship Bias?
- The red dots in the image below shows where bullet holes in WWII bombers coming back from combat were most pronounced
- The impulse was to begin adding armor to the areas of the plane with lots of holes
- Statistician Abraham Wald of Columbia University highlighted that the impulse was wrong: the planes that made it back, had made it back, and were survivors
- The “right” thing to do was reinforce the areas around the engines and cockpit from enemy fire
- Once written, does it not seem obvious that planes with bullets hitting the engines might be less likely to survive?
Source: Survivor Bias Image from Wikimedia
Survivorship Bias in Finance Fails to Make the Case for High Priced Stocks
One of our many thoughtful subscribers noted that some advocates of high-priced growth stocks believed they had identified the “next Amazon.” We have brought empirical rigor to the fallacy of this thinking in our piece Who is the Next Amazon. We also used tongue-in-cheek anecdotes to debunk this nonsense in our free post Apple II Flashback – the Fantasy of Predicting the Future.
Our friend, former fund manager and current Director of Research, Roger N. asked us, “if you looked at the 50 best-performing stocks over the last 20 years, what did they look like at the beginning?” Wise in the ways of survivorship error, Roger and the KCR team believed that this selection process would produce a crop of high-priced stocks like Amazon while omitting the other poorly performing high-priced stocks 20 years ago.
We were wrong. The average returns of the 50 best performing stocks was 5,292% over the last 20 years. The chart below shows how the market’s 50 best-performing stocks over the last 20 years looked 20 years ago:
Left Four Bars:
- If the average percentile of valuation is above the 50th percentile, it means they were cheaper than the market average
- The 50 best performers over the last 20 years had ~average starting valuations
Right Four Bars:
- If the percentile of quality and growth is above the 50th percentile, it means the factor was higher
- The 50 best performers over the last 20 years had above average starting quality (ROE & Total Yield above average) and slower than average sales growth (1, 3 Year Sales Growth less than the 50th percentile)
Even with the sample selection loaded with survivorship bias, the 50 best-performing stocks in the US market over the last 20 years began their journeys to greatness as firms with cheap to average valuations, above-average quality, and below-average growth rates.
This newsletter has been relentless in highlighting that the rising risks of a bear market and the possible increase in Enron-type fraud have made investing in high-quality stocks at reasonable prices more important, in our view. Enamored with glamour stocks, investors have dismissed our preference for insisting on some margin of safety when investing.
If you would like to read more about survivorship bias and how it impacts any analysis of mutual fund returns, the material available from the University of Chicago is a terrific resource.
Please see below a list of the 50 best-performing stocks in the US market over the last 20 years with basic starting and current valuations: