Size

Small-cap stocks earn a premium — but it has nearly vanished.

SMB (Small Minus Big)Small-cap premiumSmall firm effect

Typical IS Sharpe

0.3 – 0.6

Typical OOS Sharpe

0.0 – 0.2

Capacity

Small-cap

Signal decay

~12m half-life

Low turnoverLong-only viable

Overview

Banz (1981) first documented that small-cap stocks earn higher average returns than large-cap stocks, even after adjusting for CAPM beta. Fama and French (1992) incorporated size alongside book-to-market in their influential two-factor extension of the CAPM. For decades, the "small firm effect" anchored multi-factor models and justified small-cap investing as a distinct risk premia. However, the post-publication evidence has been disappointing: after adjusting for transaction costs, the size premium largely disappears in the US, and Hou and van Dijk (2019) show the premium has become statistically indistinguishable from zero in recent decades.

Economic Intuition

Risk-based: small firms are more exposed to business cycle risk, have higher costs of distress, and are more vulnerable to credit tightening. Liquidity risk: small stocks are illiquid and require compensation for liquidity risk beyond what CAPM captures. Mispricing: small firms receive less analyst coverage, creating opportunities for informed trading. The modern consensus is that the pure size premium is largely an artifact of extreme microcaps (bottom decile by market cap) that are impossible to trade in size. Novy-Marx and Velikov (2015) and others show transaction costs eliminate the premium for anything other than the very smallest positions.

Out-of-Sample Evidence

Weak OOS survival

Size is the canonical example of a factor that looked strong in the original sample but has failed decisively out-of-sample in its pure form. The lessons: (1) publication can accelerate arbitrage of a premium, (2) transaction costs must be modeled explicitly, (3) a large in-sample Sharpe ratio on a long-only factor may reflect illiquidity risk that is uncompensated net of costs. This is why the platform always reports gross-of-cost and net-of-cost metrics separately. The size interaction — small firms with high quality or high momentum — has better OOS survival than pure size.