How we value stocks: many models, one quality score, and why spreading works

Transcript

Hey, I'm Peter. Today we look at how a company's numbers turn into a valuation. Three questions: What is the stock worth? How good is the company? And how do you build a portfolio that does not hang on a single stock? Let's go.

Every valuation model has a blind spot. A model built on dividends cannot make sense of a company that pays none. One built on assets cannot make sense of a software company. Back in 1969, researchers Bates and Granger showed that if you combine several forecasts, the result lands closer to the truth on average than the single ones. A review of 50 years of research reached the same conclusion in 2022.

That is why the calculator runs 26 models from eight families for every stock. Some look at the money the company will earn in the future. Others compare it with similar companies in the same industry. Others again look at dividends, or at what the company owns. If a model does not fit the company, like a dividend model for a company that pays no dividend, it stays out. Better fewer voices than a wrong one.

And each model does not compute one value, but three: a cautious case for when things go worse, a middle case and a good case. Only the blend across all fitting models gives the fair value. The key point: the range is part of the answer. When the cases sit close together, the models agree. When they are far apart, the company is hard to value. And when the data is not enough, the calculator would rather show no fair value at all.

How do we know whether this works? Every ingredient is tested against history before it is allowed into the calculator. Our back-testing reaches 34 years into the past and only uses the numbers that were known at the time. We show you openly what came out of it, and what did not.

The second look goes to the company itself. The quality score from 0 to 100 combines 37 factors from ten areas, such as how much the company earns, how stable that is over the years and how solid its financing is. Why does that matter? Researchers at AQR studied stocks from the US and 24 other countries. The result: quality is only partly reflected in the price. Good companies were often the better investment relative to their risk. So a low price alone is not enough. Value and quality together give the picture.

Now to the portfolio. In 1952, a young economist named Harry Markowitz wrote a paper that later earned him the Nobel Prize. His idea: what matters is not only how risky a stock is, but how it moves together with the others. An example: each stock swings about 30 percent a year, and they only loosely move together. With ten such stocks, the swing of the portfolio drops to about 18 percent. With 30, to about 17. But even with a thousand, about 16 percent remains. That remainder is market risk. You cannot spread it away.

So how many stocks do you need? A study from 1968 said about ten, after that each extra one adds little. A study from 1987 arrived at 30 to 40. Everyone agrees on this: a single stock is too few, and ten stocks from the same industry are hardly spread at all, because they fall together. That is exactly what the diversification bar in your watchlist looks at, as a number from 0 to 100. With Pro: spread by sector, region, size and correlation, with a swap idea. The swap idea is chosen by how closely a stock moves with the ones you already hold.

In short: many models instead of one opinion, three cases instead of one number, quality next to price, and spreading instead of luck. In the next lesson, we look at how much stocks depend on the overall market and where concentration risks hide. Important: this is not financial advice.