Why is beta only the first factor?
The performance course asked whether a manager's return was skill or a rising market, and answered with beta: how much of a share's move is the market's move, scaled. This course starts where that answer runs out. Beta explains part of why shares differ from each other, and for fifty years the interesting question has been what explains the rest.
What the market model actually says
The capital asset pricing model — Sharpe in 1964, Lintner in 1965 — makes one claim: a share's expected return above cash is its beta times the market's expected return above cash. Nothing else should be paid for, because everything else can be diversified away. A share with a beta of 1.2 should earn 20% more than the market's excess return; a share with a beta of 0.8, 20% less. The portfolio-theory course's two kinds of risk is the same idea from the other side — only the risk you cannot diversify earns a premium, and beta is the measure of it.
It is a beautiful claim and it is testable, and when it was tested on decades of US data it failed in a specific way.
Where it fails
Two failures matter here. First, a single share's beta explains far less of its return than people assume: regress a large company's monthly returns on the market and the fit is often under a half. The rest is the company, its industry and noise. Second — and this is the door into the course — groups of shares sorted on things that have nothing to do with beta earned systematically different returns. Small companies earned more than large ones. Cheap companies, measured by book value against price, earned more than expensive ones. Neither difference was explained by beta; if anything, the low-beta end of some sorts earned more.
A model that says only beta is paid, facing data in which other things are paid, has two possible responses. Either those other things are risks the model forgot, or the market is mispricing them. The fourth lesson is about that choice. For now the fact is enough: beta is one factor, and it is not the only one the data pay for.
The word "factor"
A factor is a characteristic that sorts shares into groups with different average returns, measured as the return of the top group minus the bottom. The market factor is the oldest: all shares minus cash. Size is small minus big. Value is cheap minus expensive. Each is a long-short portfolio you could hold, which is what makes the return a number rather than a story.
Two things a factor is not. It is not a sector: "banks" is a group, but a bank's return is not explained by being a bank the way a small company's return is partly explained by being small across every sector. And it is not a stock pick: a factor is a rule applied to hundreds of names at once, and it works, when it works, on the average, with any single name free to disappoint.
In the data
A share's data record states its beta against the market, computed by the data provider. Apple, then Coca-Cola:
Beta is the only factor exposure a company's record states outright. Every other one in this course is computed from ordinary published figures — market capitalisation, book value, earnings, the price history — which is the point of the next unit.
Try it now
Read the beta of Apple and of Coca-Cola above. Say what the market model predicts about their relative returns in a year the market rises 10%.
Now look at what the market itself did over the window this course measures everything against, and note the scale a beta of 1.0 is multiplying:
- Reopen Was that skill, or just a rising market? and write down, in one line, what that lesson called the part of a return beta does not explain. This course is about what lives inside that line.