Factor investing is the systematic practice of targeting specific, persistent drivers of stock returns. Rather than picking individual securities, factor investors construct portfolios that tilt toward characteristics historically associated with higher risk-adjusted returns.

The central question

What explains the cross-section of expected stock returns? The CAPM (Sharpe 1964) proposed a single answer: market-beta. Stocks with higher sensitivity to the market should earn proportionally higher returns. This prediction fails empirically. Fama and French (1992) showed that beta has no explanatory power once size and book-to-market are controlled for. The search for what does explain returns produced the factor investing framework.

The major factors

Six factors have emerged from the academic literature as robust, persistent, and pervasive:

FactorConceptLongShortKey paper
RM-RFMarketStocksT-billsSharpe (1964)
SMBSizeSmall capsLarge capsFF (1993)
HMLValueHigh B/MLow B/MFF (1993)
UMDMomentumPast winnersPast losersJT (1993)
RMWProfitabilityHigh OPLow OPFF (2015)
CMAInvestmentLow investHigh investFF (2015)

Quality (QMJ) is a composite factor combining profitability, growth, and safety, developed by Asness, Frazzini, and Pedersen.

Evolution of factor models

The history of factor models is a progressive expansion of what counts as a “priced” risk dimension:

  1. CAPM (1964): 1 factor (market). Theoretically elegant but empirically rejected.
  2. Fama-French Three-Factor (1993): 3 factors (market, size, value). Captured most cross-sectional variation. Dominant benchmark for two decades.
  3. Carhart Four-Factor (1997): 4 factors (FF3 + momentum). Added the strongest anomaly. Standard for performance evaluation.
  4. Fama-French Five-Factor (2015): 5 factors (FF3 + profitability, investment). Motivated by the dividend discount model. Striking finding: HML becomes redundant when profitability and investment are included.

Commercial risk models like Bloomberg’s MAC3 use 14+ style factors for risk forecasting, going beyond the academic goal of explaining expected returns.

Key interactions and combinations

The most important structural relationship in factor investing is the negative correlation between value and momentum (~-0.6 within equity markets). This makes them natural complements:

  • A 50/50 combination across all asset classes produces a Sharpe ratio of 1.42
  • Maximum drawdown drops from -77% (momentum alone) to -30% (60/40 value/momentum)
  • Each hedges the other’s worst periods

Profitability and value are also negatively correlated (-0.57), but complementary: controlling for profitability dramatically improves value strategy performance, and vice versa.

See value-momentum-interaction for detailed evidence.

Global evidence

Factor premia are not limited to U.S. equities. Asness, Moskowitz, and Pedersen (2013) document significant value and momentum premia across:

  1. U.S., U.K., European, and Japanese individual stocks
  2. Country equity index futures
  3. Government bonds
  4. Currencies
  5. Commodity futures

The returns exhibit strong common factor structure across all these markets, suggesting global risk factors rather than market-specific anomalies.

Open debates

Risk vs. behavioral explanations: Do factor premia compensate for systematic risk (Fama-French view) or exploit persistent mispricing (behavioral view)? The evidence is mixed:

  • Profitability challenges distress-based risk stories (profitable firms are less distressed, yet earn higher returns)
  • Quality stocks appear safer, not riskier, yet earn significant alpha
  • Global common factor structure supports risk-based explanations
  • Analyst forecast errors systematically favor junk over quality stocks, supporting mispricing

Factor crowding and decay: As factor strategies become more widely adopted, their premia may diminish. Asness et al. (2014) find no degradation in momentum out-of-sample (1991-2013), but the question remains open.

HML redundancy: Is value truly redundant once profitability and investment are controlled for? Fama and French (2015) caution this may be sample-specific.

From theory to practice

The gap between academic factor models and implementable strategies involves:

  • Transaction costs: real-world costs are roughly one-tenth of academic estimates for large systematic managers (AQR data)
  • Capacity: factors work among large, liquid stocks, not just small caps
  • Risk models: commercial models like MAC3 translate academic factors into portfolio construction tools with multi-horizon risk forecasts