What a Factor Is: Risk Premia, Anomalies and Long-Short versus Long-Only

What you will learn

• What a factor is and how it differs from a sector, a theme or a stock-picking opinion.
• Why academics split factor returns into risk premia and anomalies, and why the label matters for you.
• How the Fama–French research portfolios are built and what their long-short returns actually measure.
• Why a long-only factor ETF usually captures roughly half of a long-short factor spread, and less after costs.

Explanation

A factor is a measurable characteristic of a share that has been linked, in long historical samples, to differences in average return or risk across many stocks. A factor is not a story about one company; it is a rule that sorts thousands of shares into groups and compares the groups.

The classic method comes from Eugene Fama and Kenneth French. In their 1992 and 1993 papers they sorted US shares by market capitalisation and by book-to-market (book value of equity divided by market value) and formed portfolios. The HML factor ("high minus low") is the return of the cheap (high book-to-market) portfolios minus the return of the expensive ones. SMB ("small minus big") is small-company portfolios minus large-company portfolios. Later work added momentum (Carhart, 1997), and in 2015 Fama and French added profitability (RMW, robust minus weak) and investment (CMA, conservative minus aggressive). The market factor itself, Mkt-RF, is the market return minus the risk-free rate.

Two features are easy to miss. First, these factors are long-short: one unit means buying one euro of the "good" side and selling short one euro of the "bad" side. Second, they are self-financing on paper: the short sale pays for the long position, so the factor return is a spread, not a return on invested capital. Nobody can hold HML directly; the series excludes borrowing fees, trading costs and the practical limits of short selling.

Why does a factor earn anything? There are two families of explanation (lesson 11 goes deeper). A risk premium is extra expected return for carrying a risk other investors want to avoid, such as owning distressed companies in a recession; such a premium should persist because it is payment for discomfort. An anomaly is a return pattern that cannot be explained by risk and is usually attributed to behavioural mistakes or to constraints that stop arbitrage. An anomaly can shrink when enough capital chases it. Most factors are probably a mix, and the historical observation is not a guarantee of a future premium.

The product you can actually buy is a long-only factor ETF. It owns shares that score well on the factor and nothing on the short side. Its return is the market return plus a partial factor spread, not the spread itself. The calculation below shows why "partial" typically means about half.

Worked example

Hypothetical example: imagine a market split into two equal halves by a value score. Over one year the cheap half returns 10 percent and the expensive half returns 7 percent. All figures are before fees, spreads and taxes.

• Long-short factor spread: 10% − 7% = 3 percentage points.
• Market return (equal halves): (10% + 7%) ÷ 2 = 8.5%.
• A long-only fund holding the cheap half returns 10%, which is 1.5 percentage points above the market, exactly half of the spread.

Now assume a real long-short implementation with a hypothetical borrowing fee of 0.5% per year on the short leg and trading costs of 0.4% per year. The realised spread falls to 3% − 0.5% − 0.4% = 2.1 percentage points. The long-only fund avoids the borrowing fee but gives up the short-side return: it captures 1.5 points of a 3-point spread and keeps full market risk. If the market falls 30%, the fund falls by roughly 30% too, plus or minus its tilt.

So when a paper reports that a factor "earned 4 percent per year", a long-only product built on it should be expected to add far less than 4 points above a market index, before costs, and nothing is promised.

Common mistake

Practice activity

On paper, estimate the expected excess return of a hypothetical long-only value fund. Assume the fund has a market beta of 1.0, a loading of 0.35 on the HML factor, an expected market excess return of 6% per year and an expected HML spread of 2.5% per year. Ignore fees. (A loading of 0.35 means the fund moves 0.35 units for each unit of HML; lesson 9 shows how to estimate it.) Calculate the total expected excess return and the part from the value tilt, then state what happens if the HML spread turns out to be zero.

Key takeaways

• A factor is a sorting rule across many stocks, tested with long-short portfolios such as HML, SMB, RMW and CMA.
• Academic factor returns are self-financing spreads before costs; they are not returns you can buy.
• A long-only product typically captures around half of a factor spread and keeps full market risk.
• Risk-premium and anomaly explanations imply different persistence; historical evidence is not a promise.

Next step

The next lesson examines the value factor in depth: how it is defined, why its definition matters and what happened between 2007 and 2020.