Factor investing targets specific drivers of return that have been identified through decades of academic research. The most established factors include value (cheap stocks outperform), momentum (recent winners continue winning), quality (profitable companies outperform), size (small stocks outperform), and low volatility (less volatile stocks deliver higher risk-adjusted returns).
Multi-factor models combine several factors to build diversified portfolios that are less dependent on any single return driver. Professional quant funds continuously research new factors and optimize their combinations.
A factor, in plain terms, is a shared characteristic that seems to explain why groups of stocks earn what they earn. Instead of asking whether one company is a good buy, factor investors ask a broader question: do stocks with this trait, as a group, tend to earn more than stocks without it? That shift — from stories about companies to statistics about traits — is the heart of quantitative investing.
Value means buying stocks that are cheap relative to their fundamentals — price compared to earnings, book value, or cash flow. The idea: markets overreact to bad news, leaving solid businesses on sale. Value can underperform for painfully long stretches; it lagged badly for most of the 2010s.
Momentum means buying stocks that have risen over roughly the past year and avoiding recent losers. It sounds too simple to work, yet it's one of the most persistent patterns researchers have found. Its dark side is the crash: when markets snap back sharply, momentum portfolios can lose years of gains in weeks.
Quality means favoring companies with high profitability, stable earnings, and low debt. Boring, sturdy businesses have historically earned more than their riskiness suggests they should.
Low volatility is the strangest one. Basic theory says risky stocks should pay more. In practice, the calmest stocks have delivered better risk-adjusted returns than the wildest ones. One popular explanation: investors overpay for lottery-like stocks, chasing big wins that mostly never come.
It's worth being precise about what the research says, because it's easy to oversell.
The strong version of the evidence: value, momentum, quality, and low volatility have shown up across many decades, many countries, and many asset classes. Fama and French put value and size on the academic map in 1992; Jegadeesh and Titman documented momentum in 1993. Patterns that appear in Japanese stocks, European bonds, and commodity futures alike are less likely to be pure statistical flukes.
The honest caveats: size, once a headline factor, looks much weaker in modern data. Every factor has endured drawdowns lasting five to ten years — long enough to shake out almost anyone. And a large replication literature has found that many published factor results shrink, sometimes by half, when re-tested on fresh data.
So the fair summary isn't that factors guarantee outperformance. It's that a handful of traits have historically tilted the odds, unreliably, over long horizons — and holding on through the bad stretches is the real price of admission.
Statistical arbitrage (stat arb) uses mathematical models to identify mispricings between related securities. The classic example is pairs trading: finding two historically correlated stocks that have temporarily diverged, going long the underperformer and short the outperformer, and waiting for convergence.
Modern stat arb strategies operate across thousands of securities simultaneously, use complex cointegration and machine learning models, and require sophisticated risk management. These strategies typically generate many small gains, relying on the law of large numbers for consistent profitability.
The catch is in the word statistical. Nothing forces two diverged stocks to converge. Sometimes they've split apart for a real reason — one company's business genuinely changed — and the gap never closes. A stat arb book survives by sizing each bet small, so no single broken pair can sink the whole portfolio.
Here's the uncomfortable dynamic at the center of quant investing: publishing an edge starts destroying it.
Once a factor appears in a journal, funds launch products around it. Their combined buying pushes up the prices of the stocks the factor favors, which shrinks the future returns the factor was supposed to deliver. Researchers McLean and Pontiff studied nearly a hundred published anomalies and found returns fell substantially after publication — a decent chunk of the paper edge simply evaporated.
Crowding adds a second risk beyond slow decay. When many funds hold the same factor portfolio and one of them has to sell fast, they all bleed together. August 2007's quant meltdown worked exactly this way: stat arb funds with overlapping positions unwound at once, and strategies that looked uncorrelated on paper crashed in unison.
The lesson for backtesters is humbling. A pattern discovered in decades-old data was competed over for years before you found it. Assume the live edge is smaller than the historical one — sometimes much smaller.
You don't need a PhD to put factor thinking to work — you need patience and a tolerance for looking wrong.
Start by understanding what tilts your portfolio already has. A tech-heavy growth portfolio is, in factor terms, short value and long momentum, whether you meant it that way or not. Naming your exposures is half the benefit.
If you build factor strategies, backtest them across full market cycles, including 2000 to 2002, 2008, and 2022 — the stretches that punish specific factors. Watch how factors behave together: value and momentum tend to offset each other's bad years, which is why combining them has historically been steadier than holding either alone.
And decide, before you start, how long you'll tolerate underperformance. The historical answer for factor investors is brutal: years, not months. Most people who abandon a factor do it at the bottom of its cycle. The evidence can be real and still do you no good if you can't hold on through the stretch when it looks broken.
Previous: Algorithmic Trading · Next: Portfolio-Level Backtesting · Financial glossary
← Back to all investing concepts