Factor Models in Equity Trading: What Momentum, Value, and Quality Actually Mean in Practice

Most introductions to factor models in equity trading spend three paragraphs explaining that "momentum is the tendency of recent winners to keep winning" and then hand you a backtest chart that ends in 2019. That framing skips the parts that actually matter for anyone who tries to trade these signals: how the factors are constructed in practice, why the academic returns don't show up in live portfolios, and which factors have held up well enough to be worth building around. This is the version that doesn't skip those parts.

What a Factor Actually Is (And What It Isn't)

In quantitative equity, a "factor" is a systematic characteristic of a stock that has historically predicted its future returns in the cross-section — meaning it ranks stocks against each other, not against an absolute benchmark. The Fama-French research group has catalogued the most durable ones, and their data library is public. But the critical thing to understand upfront is this: academic factors are constructed as long-short portfolios, typically buying the top decile of some ranked universe and shorting the bottom decile, rebalanced monthly, with no transaction costs assumed. That's not a trading strategy. It's a measurement tool. The gap between factor returns in academic papers and what you can actually capture in a long-only portfolio with real friction is where most retail exposure to "factor investing" lives — and it's a significant gap.

A 2023 review of cross-sectional return research (Harvey, Liu, and Zhu) catalogued over 316 published factors in finance journals. The uncomfortable implication: with that many factors tested on overlapping datasets, many published findings are false positives. The factors that have survived multiple out-of-sample tests, multiple international markets, and multiple decades of scrutiny can be counted on one hand. Momentum, value, and quality are three of them.

Momentum: The Factor That Shouldn't Exist But Does

Momentum is the most academically embarrassing of the major factors — Fama himself called it the "main embarrassment" of the three-factor model when it refused to be explained away. The core finding from Jegadeesh and Titman (1993) is simple: stocks that outperformed the market over the past 12 months (excluding the most recent month to avoid the short-term reversal effect) continue to outperform over the following 3–12 months, and vice versa for underperformers. This is sometimes called the 12-1 momentum signal.

How the Signal Is Constructed

The standard construction: for each stock in a universe, calculate its 12-month return excluding the most recent month (so month t-12 to month t-2). Rank stocks within your universe by this figure. Long the top quintile or decile, short the bottom. In the Fama-French data running from 1927 to 2023, the UMD (Up Minus Down) momentum factor delivered roughly 7–9% annualized excess returns with a Sharpe ratio around 0.5–0.6 — high for a single long-short factor.

The Part the Backtest Hides: Momentum Crashes

Here is what the summary statistics conceal. Momentum returns are negatively skewed. The strategy generates steady positive returns most of the time and then suffers abrupt, violent reversals during market panic recoveries. The two worst months for a momentum strategy in the U.S. equity sample were July and August of 1932: the past-loser decile returned 232% over those two months while the past-winner decile fell sharply. More recently, the March–June 2009 rebound after the financial crisis crushed momentum strategies — the past losers (banks, cyclicals) exploded higher while the past winners (defensives) lagged badly. Daniel and Moskowitz (2016) formalized this pattern: momentum crashes occur in "panic" states following significant market declines, when volatility is elevated and the market is rebounding. Crucially, these states are partially forecastable in advance.

The practical implication: a raw momentum strategy has fat left-tail risk. Institutions running momentum typically apply a volatility scaling overlay — reducing position size when realized or implied market volatility is elevated. Without this, the raw factor Sharpe ratio significantly overstates what a live portfolio achieves.

Transaction Costs and Turnover

A monthly-rebalanced large-cap momentum strategy turns over roughly 50–80% of the portfolio per year. In liquid large-cap equities with tight spreads, this is manageable. Mid-cap and small-cap momentum — which shows stronger academic returns — runs into much higher transaction costs and market impact that erode the premium substantially. AQR's research suggests net-of-cost momentum returns in large caps remain positive but are roughly half the gross returns. For an individual investor in a taxable account, the turnover also generates short-term capital gains, which further compresses after-tax returns in ways that don't appear in any academic paper.

Value: The Factor That Almost Died and Might Have a Point

The value factor — HML (High Minus Low) in the Fama-French framework — buys stocks with high book-to-market ratios ("cheap" on a fundamental basis) and shorts stocks with low book-to-market ratios ("expensive" growth stocks). It was the foundational factor of the original Fama-French three-factor model in 1992 and delivered strong positive returns from the 1960s through the early 2000s.

What Happened to Value in the 2010s

From 2007 to mid-2020, the HML factor sustained a drawdown of approximately 55% — one of the longest and deepest factor drawdowns ever recorded for a major premium. Growth stocks (large-cap tech especially) dominated. The natural question is whether value is dead as an effect.

The most rigorous analysis (Arnott, Harvey, Rattray, and Sammon, Financial Analysts Journal 2021) attributes the drawdown to two separable causes. First, the traditional book-to-market measure systematically misvalues technology companies because it ignores the value of intangible assets — things like brand, software, and R&D investments that appear as expenses rather than assets on a GAAP balance sheet. A company spending $5 billion per year on R&D is not a cheap stock on a book-value basis, but it may have enormous economic value. When you capitalize intangibles, the adjusted value factor substantially outperforms the raw book-to-market version. Second, value stocks got genuinely cheap on a relative basis starting around 2007 — the valuation spread between growth and value portfolios reached historically extreme levels by 2020, at the bottom percentile of the historical distribution. That spread compression, not a structural death of the premium, explains most of the drawdown. Both signals — intangible adjustment and valuation spread — were pointing toward eventual mean reversion, not permanent impairment.

What Value Actually Requires to Work

Value is a slow factor. Academic research uses annual rebalancing. The premium tends to compound over 3–5 year horizons, not months. Running a value factor portfolio at monthly rebalancing destroys much of the edge in transaction costs while providing little benefit — the signal doesn't move that fast. Value also works better in small caps, where stocks are less efficiently priced and analyst coverage is sparse. But small-cap value investing at retail scale means accepting illiquidity, wider spreads, and the real risk of market impact in smaller names.

Quality: The Most Underappreciated Factor

Quality is the newest of the three to be formalized academically, and it tends to get less attention because it doesn't have a dramatic story — it doesn't crash violently like momentum or stage a decade-long drawdown debate like value. But it has arguably the most consistent real-world evidence.

How Quality Is Measured

There's no single agreed definition of quality, which is part of why it's underrepresented in index-fund marketing. The main dimensions are:

• Profitability: Robert Novy-Marx's 2013 paper showed that gross profitability — gross profit divided by total assets — predicts cross-sectional stock returns with roughly the same power as the book-to-market ratio. High gross profitability firms outperform low gross profitability firms persistently. This became the profitability factor (RMW, Robust Minus Weak) in the Fama-French five-factor model in 2015.
• Earnings stability: Companies with consistent, predictable earnings outperform those with volatile or negative earnings. Metrics include the standard deviation of year-over-year earnings growth and the frequency of negative earnings quarters.
• Balance sheet health: Low leverage, high interest coverage, and strong liquidity ratios predict outperformance. Piotroski's F-Score, a composite of nine binary signals across profitability, leverage, and operating efficiency, was an early systematic quality measure and still holds up well in research.

Asness, Frazzini, and Pedersen's Quality Minus Junk (QMJ) factor — a composite quality measure covering profitability, growth, and safety — has delivered positive returns in all 24 countries studied in their research. That's the kind of out-of-sample robustness that's genuinely rare in factor research.

Why Quality Works and Keeps Working

Unlike momentum, quality doesn't crash. Unlike value, it doesn't require a multi-year patience horizon to realize. High-quality companies are expensive relative to book value — the market isn't stupid about them — but they're not expensive enough given their future earnings power. Investors systematically underestimate the persistence of corporate profitability. High-ROE companies stay high-ROE longer than most models assume; low-quality companies mean-revert toward insolvency faster than most investors price in. This behavioral mispricing, rather than risk compensation, explains much of the quality premium — which means it's less likely to be eliminated by rational arbitrage alone.

Quality also interacts well with value. A stock that is cheap and high quality is rare and tends to be significantly undervalued. This combination — sometimes called "quality value" or captured by metrics like free cash flow yield — has historically produced the strongest risk-adjusted returns in the factor literature.

The Practical Problem: What You Can Actually Do With This

Factor ETFs: Real Exposure vs Marketing Exposure

If you're not running a quant equity strategy yourself, the most accessible factor exposure comes through ETFs. But factor ETF labels are deceptive. BlackRock's MTUM doesn't run a standard 12-1 momentum strategy — it uses 6-month and 12-month return momentum, holds semi-annually, and applies a volatility adjustment. VLUE screens for value on four metrics simultaneously. QUAL targets MSCI's composite quality score. These are real factor exposures, but they're diluted by diversification, reconstitution timing, and index construction choices that prioritize investability over factor purity.

A better-constructed five-factor portfolio for a long-only investor might look like this, based on factor loading analysis: 60% VTI for broad market exposure, 15% AVUV (Avantis U.S. Small Cap Value) which loads strongly on size, value, and profitability simultaneously, 10% COWZ (cash flow yield as a value/investment factor proxy), and 10% QUAL for explicit profitability tilt. Blended expense ratio runs around 0.12%. This loads positively on all five Fama-French factors in regression analysis, which you can verify through the AQR or Portfolio Visualizer factor regression tools.

Doing It Yourself: The Minimum Bar

Running your own factor strategy on individual stocks is a different undertaking. The minimum viable setup requires: a reliable fundamental data source (Compustat via WRDS, Sharadar, or similar), the ability to rank a universe of at least 500–1000 stocks monthly, and realistic transaction cost modeling before you trust any backtest. If your backtest doesn't include at least 0.05–0.10% round-trip costs for mid/large caps and 0.15–0.30% for small caps, plus market impact assumptions for position sizes above $50K per name, the results are unreliable.

The other reality: factor strategies require patience horizons that most individuals don't have. A value factor tilt can underperform for 3–5 years during a growth cycle. A momentum strategy can blow up 20–30% in a single month during a panic recovery. The academic returns assume a mechanical, emotionless rebalancing that ignores drawdowns. In practice, almost no individual investor sticks with a systematic strategy through a 30% drawdown. That behavioral gap is the largest reason factor premiums haven't been arbitraged away — most people who try to capture them quit before capture.

Where the Real Edge Sits

Based on the out-of-sample evidence, the most durable factor positions for a long-only equity investor are: a quality tilt (high gross profitability, clean balance sheet, stable earnings) combined with a value screen applied within industries rather than across them (industry-adjusted value avoids the problem of comparing banks to software companies on book-to-market). Momentum is more valuable as a timing signal than a long-term tilt — using momentum to decide when to add to a quality-value position rather than to build a pure momentum portfolio sidesteps most of the crash risk while preserving the informational content of recent price action.

The factor zoo problem is real: of the 316+ factors published in academic journals, most won't survive transaction costs, out-of-sample periods, or a genuinely updated t-statistic threshold. Momentum, value (adjusted for intangibles), and quality have cleared that bar repeatedly over multiple decades and multiple geographies. That doesn't mean they're free money — it means they're structural tilts with a real economic rationale, measurable premia, and known failure modes. Understanding the failure modes is as important as understanding the premia.

Practical Takeaway

If you're approaching factor models as an active trader, the three things most articles don't tell you: first, the academic long-short factor returns are not your returns — they're a signal purity benchmark that nets out market beta and ignores costs. Second, the factors that work best in combination are quality and value (complementary, uncorrelated failure modes), while momentum works best as a timing overlay rather than a standalone tilt. Third, factor premiums require multi-year time horizons and mechanical discipline to capture — if you're likely to override your strategy based on 6-month performance, you're not capturing the premium, you're just adding tracking error. The edge in factors is real. The ability to systematically harvest it is far rarer than the marketing suggests.