The Carhart Four-Factor Model: Adding Momentum to Asset Pricing

The Debate That Shook Wall Street

In the early 1990s, the mutual fund industry was riding a wave of unprecedented growth. Billions of dollars poured into actively managed funds, their marketing departments trumpeting the star managers whose track records promised consistent outperformance. Financial magazines published annual rankings of top-performing funds, and investors chased last year's winners with an almost religious fervor. The implicit promise was seductive: skilled managers could reliably beat the market, and past performance was the key to identifying them.

But beneath the glossy advertisements, a quiet academic war was raging. On one side stood practitioners and some researchers who pointed to evidence of "hot hands" in mutual fund management, performance that persisted from year to year in ways that seemed to defy pure chance. On the other side, efficient market proponents argued that apparent persistence was a statistical mirage, the inevitable result of thousands of managers flipping coins, with the lucky ones mistaken for geniuses. The stakes were enormous: if persistence was real and identifiable, the entire case for passive investing would weaken. If it was illusory, the active management industry was charging premium fees for a service it could not deliver.

Into this charged debate stepped Mark Carhart, then a doctoral student at the University of Chicago. His 1997 paper, "On Persistence in Mutual Fund Performance," published in the Journal of Finance (Carhart, 1997), would not merely adjudicate the dispute. It would introduce a tool, the four-factor model, that would become one of the most widely used benchmarks in empirical finance for decades to come.

The Intellectual Foundation

To understand Carhart's contribution, one must first appreciate the models that preceded it. The Capital Asset Pricing Model (CAPM), developed independently by Sharpe (1964), Lintner (1965), and Mossin (1966), proposed that a single factor, the excess return of the market, could explain the cross-section of expected returns. Under the CAPM, a fund's risk-adjusted performance, its alpha, was measured as the intercept in a regression of fund excess returns on market excess returns. Any positive alpha represented genuine skill.

Yet by the early 1990s, the CAPM's shortcomings were well documented. Fama and French (1993) demonstrated that two additional factors, one capturing the return spread between small and large stocks (SMB, or Small Minus Big) and another capturing the spread between high book-to-market and low book-to-market stocks (HML, or High Minus Low), substantially improved the model's explanatory power. Their three-factor model became the new standard for academic research.

However, even the three-factor model left a significant anomaly unaddressed. Jegadeesh and Titman (1993) had documented a powerful pattern: stocks that had performed well over the previous three to twelve months tended to continue performing well, while recent losers continued to lose (Jegadeesh and Titman, 1993). This momentum effect was large, persistent across time periods, and could not be explained by the Fama-French factors. It was, in the language of asset pricing, an anomaly in search of a model.

Carhart's Methodological Innovation

Carhart's insight was to marry two separate research agendas. He recognized that the debate over mutual fund persistence and the momentum anomaly might be deeply connected. If some mutual funds happened to hold recent winners, simply by following momentum-like strategies or by coincidence, their short-term outperformance would look like skill. But it would actually be compensation for bearing a systematic risk factor that the existing models failed to capture.

To test this hypothesis, Carhart constructed what is now known as the four-factor model. It extends the Fama-French three-factor model by adding a fourth factor: WML (Winners Minus Losers), also sometimes called UMD (Up Minus Down) or PR1YR in Carhart's original notation. This factor captures the return of a portfolio that is long stocks with high prior-year returns and short stocks with low prior-year returns.

The model takes the following form:

R_i - R_f = alpha_i + beta_1(R_m - R_f) + beta_2(SMB) + beta_3(HML) + beta_4(WML) + epsilon_i

Where R_i is the return of fund i, R_f is the risk-free rate, R_m is the market return, and the four factors capture exposure to market risk, size, value, and momentum respectively. The intercept, alpha, now represents the portion of return that cannot be explained by any of these four systematic factors, a much stricter test of genuine manager skill than previous models provided.

The Dataset and Findings

Carhart assembled a comprehensive dataset of 1,892 diversified equity mutual funds spanning January 1962 through December 1993, one of the most extensive samples used in fund performance research at that time. He sorted funds into decile portfolios based on their prior-year returns and tracked their subsequent performance, a methodology designed to directly test whether past winners continued to win.

The Persistence Puzzle, Largely Solved

The results were striking. When evaluated using the CAPM alone, there appeared to be meaningful persistence in fund performance. Top-decile funds seemed to continue outperforming, and bottom-decile funds continued underperforming. This finding aligned with earlier work by Hendricks, Patel, and Zeckhauser (1993), who had coined the term "hot hands" to describe the phenomenon.

But when Carhart applied his four-factor model, the picture changed dramatically. The apparent persistence in the top deciles was almost entirely absorbed by the momentum factor. Funds that had performed well in the prior year tended to hold stocks with high recent returns, and it was this mechanical momentum exposure, not manager skill, that drove their subsequent outperformance. The four-factor alpha of the top-decile portfolio was economically small and statistically insignificant.

The Bottom Decile Exception

One form of persistence did survive the four-factor adjustment, but it offered no comfort to active management advocates. Funds in the bottom decile continued to perform poorly even after accounting for all four factors. Carhart traced this persistent underperformance not to negative skill, the ability to systematically pick losing stocks, but to more mundane causes: high expense ratios and excessive transaction costs. The worst-performing funds were simply the most expensive to own.

The Cost of Chasing Performance

Carhart also documented that investors' tendency to chase past performance was costly. Net of all factors, the strategy of buying last year's top-performing funds and selling last year's worst performers generated negligible risk-adjusted returns. After accounting for the transaction costs involved in annual portfolio rebalancing, the strategy was a net loser. The apparent "smart money" effect, whereby capital flows seemed to anticipate future performance, was an artifact of momentum, not investor foresight.

Relationship to the Fama-French Framework

The Carhart model occupies an interesting position in the history of factor models. It was born from a specific empirical question about fund persistence rather than from a theoretical asset pricing framework. Fama and French themselves were initially reluctant to include momentum in their models, viewing it as an empirical regularity without a clear risk-based explanation. This reluctance persisted through their development of the five-factor model in 2015, which added profitability and investment factors but deliberately excluded momentum.

Yet in practice, the Carhart four-factor model became ubiquitous. Its simplicity and explanatory power made it the default benchmark for evaluating professional fund managers, for testing trading strategies, and for academic studies of market efficiency. When researchers needed to determine whether a fund or strategy generated genuine alpha, the four-factor model was, and in many contexts remains, the first tool they reached for.

The tension between the Carhart and Fama-French approaches reflects a deeper philosophical divide in finance. Is a factor legitimate only if it can be justified by a risk-based theory? Or is empirical pervasiveness sufficient? Momentum occupies an uncomfortable middle ground. It is one of the most robust return patterns ever documented, replicated across markets, asset classes, and time periods (Asness, Moskowitz, and Pedersen, 2013). Yet no consensus risk-based explanation has emerged. Behavioral explanations, anchored in investor underreaction and delayed overreaction, remain the most popular accounts, but the debate is far from settled.

Practical Implications for Fund Evaluation

Carhart's work carries several direct implications for anyone evaluating investment managers or constructing portfolios.

Decomposing Returns

The four-factor model provides a systematic way to decompose a fund's returns into compensated factor exposures and residual alpha. A fund that appears to outperform the market by 3 percent annually might, upon four-factor analysis, reveal that 1 percent comes from a small-cap tilt (SMB exposure), 1 percent from a value tilt (HML exposure), 0.5 percent from momentum loading (WML exposure), and only 0.5 percent from genuine alpha. Since the factor exposures can be replicated cheaply through index funds or ETFs, only the alpha component justifies active management fees.

The Persistence Warning

The paper's central finding, that apparent performance persistence is mostly a momentum artifact, serves as a caution against the common investor behavior of chasing hot funds. What looks like a manager's skill is frequently the mechanical result of holding recent winners. When momentum reverses, as it periodically does with devastating speed, the "skilled" manager's outperformance evaporates.

Expense Ratios Matter

Carhart's evidence that the strongest form of genuine persistence is negative, driven by high costs, reinforces one of the most robust findings in investment research: fees are the single best predictor of future fund performance. Funds with the highest expense ratios consistently deliver the worst outcomes for investors, a finding that has been replicated numerous times in the decades since.

Criticisms and Limitations

No model is without its critics, and the Carhart four-factor model has faced several challenges over the years.

Some researchers have argued that the model's treatment of momentum as a static factor fails to capture the time-varying nature of momentum returns. Momentum strategies exhibit periods of spectacular crashes, most notably in 2009 when the long-short momentum portfolio suffered losses exceeding 40 percent in a matter of months. A static factor loading may not adequately capture a fund's dynamic exposure to this risk.

Others have noted that the model, like all factor models, is subject to the joint hypothesis problem: a finding of zero alpha could mean either that the manager lacks skill or that the model itself is misspecified. If the four factors do not fully capture all systematic risks, some genuine alpha may be incorrectly attributed to factor exposures, or vice versa.

The rise of the Fama-French five-factor model and more recent multifactor specifications has also raised questions about whether four factors are sufficient. With the addition of profitability (RMW) and investment (CMA) factors, some patterns that appear as alpha under the four-factor model may be reclassified as factor exposure under richer specifications.

Legacy and Continuing Influence

Nearly three decades after its publication, Carhart's paper remains one of the most cited works in financial economics. Google Scholar records tens of thousands of citations, a testament to the model's practical utility and intellectual influence.

The paper's impact extends well beyond academia. The four-factor model has become embedded in the infrastructure of the investment industry. Morningstar, the dominant provider of fund analytics, uses factor-based analysis rooted in Carhart's framework. Pension fund consultants employ it to evaluate manager performance. Hedge fund researchers use it as a baseline for assessing alpha generation.

Perhaps most importantly, Carhart's work contributed to a broader shift in how investors think about returns. The idea that returns should be decomposed into factor exposures and residual alpha, and that only the latter justifies active fees, is now so widely accepted that it feels obvious. But in 1997, it was a powerful and controversial claim. By demonstrating that momentum explained much of the persistence that had been attributed to skill, Carhart helped establish the intellectual foundation for the passive investing revolution that followed.

For today's investor, the lesson remains clear: before attributing a fund manager's performance to skill, ask whether the same returns could have been achieved through simple, low-cost factor exposures. The Carhart four-factor model provides the framework to answer that question rigorously.

The Fama-French Five-Factor Model Explained

Models & Frameworks12 min

Factor Timing: Can You Time the Factors?

Behavioral Finance & Timing13 min

The Momentum Factor: Why Winners Keep Winning

Factor Investing12 min

Alternative Risk Premia: Harvesting Returns Beyond Traditional Assets