Factor Investing

Factor investing is a systematic investment approach that selects securities based on specific measurable characteristics — called factors — that have been shown to drive returns. Instead of trying to pick individual winning stocks (traditional stock-picking), factor investors buy baskets of stocks that share a desirable attribute and short baskets that don't.

The intellectual origins lie in the work of Eugene Fama and Kenneth French, who showed in 1993 that the CAPM (which uses only market beta to explain returns) is incomplete. Their three-factor model added two new factors — size and value — and explained a large portion of returns that CAPM could not. This was later extended to the five-factor model (adding profitability and investment patterns) and continues to evolve.

Factor investing now represents trillions of dollars in assets, spanning both "smart beta" ETFs (retail-accessible factor strategies) and sophisticated multi-factor strategies at quant hedge funds like Citadel, Two Sigma, and AQR Capital.

Academic research has identified several factors that have historically generated excess returns:

• Value: Buy cheap stocks (low price-to-book, price-to-earnings) and short expensive ones. The value premium has been documented across markets and asset classes since the 1970s, though it famously underperformed from 2010-2020.
• Momentum: Buy recent winners (stocks with high returns over the past 3-12 months) and short recent losers. Momentum is one of the most robust anomalies in finance, documented in equities, bonds, currencies, and commodities.
• Size: Buy small-cap stocks and short large-cap stocks. Small stocks have historically earned higher returns, though the size premium has weakened in recent decades.
• Quality: Buy profitable, stable, well-managed companies (high ROE, low debt, consistent earnings) and short low-quality firms.
• Low Volatility: Buy low-volatility stocks and short high-volatility stocks. Counterintuitively, low-volatility stocks have earned higher risk-adjusted returns than high-volatility stocks — the "low-volatility anomaly."
• Carry: Buy high-yielding assets and short low-yielding assets. Most prominent in currencies and fixed income.

Each factor has a theoretical explanation (risk-based or behavioral), a long historical track record, and periods of underperformance that make it risky to bet on any single factor.

Let's construct a simple two-factor strategy combining value and momentum:

Universe: 500 large-cap U.S. stocks.

Step 1 — Score each stock on value: Rank all stocks by price-to-earnings ratio. The cheapest 20% get a value score of 5; the most expensive 20% get a value score of 1.

Step 2 — Score each stock on momentum: Rank all stocks by 12-month trailing return (excluding the most recent month). The top 20% get a momentum score of 5; the bottom 20% get a 1.

Step 3 — Combine: Composite score = 0.5 × value score + 0.5 × momentum score.

Step 4 — Portfolio construction: Go long the top 50 stocks (highest composite score) and short the bottom 50 stocks. Dollar-neutral and sector-neutral to isolate the factor exposure from market and sector risk.

Historical results (hypothetical, 2000-2024):

• Annual excess return: 5-8% (depending on implementation)
• Sharpe ratio: 0.6-1.0
• Max drawdown: -25% to -35%
• Correlation with the market: near zero (by construction)

The strategy works because it harvests two distinct return premiums. Combining factors that have low correlation to each other (value and momentum are slightly negatively correlated) improves the Sharpe ratio through diversification.

Modern quant hedge funds have taken factor investing far beyond the simple academic factors:

• Dozens of factors: Top firms use 50-200+ signals, including traditional factors (value, momentum) and proprietary signals derived from alternative data (satellite imagery, credit card transactions, web traffic, patent filings).
• Machine learning: ML models (gradient-boosted trees, neural networks) combine factors nonlinearly, potentially capturing interactions that simple linear models miss.
• Dynamic weighting: Factor exposures are adjusted based on market conditions. Value exposure might be increased during economic recoveries and reduced during recessions.
• Transaction cost optimization: Sophisticated portfolio optimization minimizes turnover and trading costs while maintaining desired factor exposures.
• Risk management: Factor risk models decompose portfolio risk into contributions from each factor, enabling precise risk control. This is essential for managing the Sharpe ratio and limiting drawdowns.

The distinction between alpha and beta has blurred: what was considered "alpha" 20 years ago (buying cheap, high-momentum stocks) is now understood as "factor beta" that can be accessed cheaply through ETFs. True alpha at top quant firms comes from proprietary signals, better execution, superior risk management, and alternative data that others haven't yet exploited.