Momentum Investing Strategy: The Complete Academic Guide to Buying Winners

Momentum — the tendency of past winners to keep winning and past losers to keep losing — is one of the strongest and most persistent anomalies in financial markets. Three decades of academic research, from Jegadeesh & Titman's foundational 1993 paper to modern global evidence, have confirmed its existence across asset classes, time periods, and geographies.

1. The Anomaly That Wouldn't Die

In the efficient markets framework that dominated academic finance through the 1970s and 1980s, a simple trading rule based on past returns should not generate abnormal profits. Prices, the theory held, already incorporated all available information. Yet momentum investing — a momentum trading strategy that systematically buys recent winners and sells recent losers — has stubbornly produced positive returns for over a century of market data.

The magnitude of the effect is striking. A zero-cost long-short momentum investing strategy in US equities has historically earned somewhere between 6% and 12% per year, depending on the exact formation and holding periods, the sample window, and how the portfolios are constructed. These returns are not explained by the Capital Asset Pricing Model, the Fama-French three-factor model, or most subsequent multifactor specifications. Even Eugene Fama, the father of efficient markets, has acknowledged momentum as the “premier anomaly.”

This article provides a comprehensive, research-grounded guide to momentum investing — how it was discovered, why it works, how to implement it, and the severe risks that have destroyed momentum traders who ignored tail events.

2. The Academic Discovery: Jegadeesh & Titman (1993)

The modern academic study of momentum begins with Narasimhan Jegadeesh and Sheridan Titman's seminal 1993 paper, “Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency,” published in the Journal of Finance, Vol. 48, No. 1, pp. 65–91.

Jegadeesh and Titman examined all NYSE and AMEX stocks over the period 1965 to 1989. They tested a matrix of strategies with formation periods of 3, 6, 9, and 12 months and holding periods of 3, 6, 9, and 12 months — sixteen combinations in total. Each month, stocks were ranked by their cumulative return over the formation period, sorted into decile portfolios, and held for the corresponding holding period.

The results were unambiguous. The strategy that bought the top decile (past winners) and shorted the bottom decile (past losers) earned an average return of approximately 1% per month (roughly 12% annualized) across most formation-holding period combinations. The 12-month formation / 3-month holding period variant was particularly strong. All sixteen strategies produced statistically significant positive returns.

Crucially, Jegadeesh and Titman also documented that the profits from momentum were not simply compensation for bearing systematic risk. The strategies produced positive alphas after controlling for market beta, size, and book-to-market. They also found that momentum profits partially reversed over longer horizons (12 to 36 months after portfolio formation), suggesting that at least some of the effect was driven by delayed overreaction rather than permanent mispricing.

3. The 12-2 Month Formation: Why Skip the Most Recent Month?

A critical implementation detail in the Jegadeesh Titman momentum strategy — one that significantly affects performance — is the practice of skipping the most recent month when computing momentum rankings. This “12-2” or “12-minus-1” formation has become the standard specification in academic research and factor construction.

The reason is the short-term reversal effect. Over horizons of one week to one month, stock returns exhibit negative autocorrelation — stocks that rose last week tend to fall this week, and vice versa. This reversal was documented by Jegadeesh (1990) and Lehmann (1990) and is attributed to several microstructure mechanisms:

• Bid-ask bounce: Closing prices alternate between the bid and ask, creating artificial negative autocorrelation in returns.
• Liquidity provision: Market makers and short-term traders earn a premium for absorbing temporary order imbalances, which reverses quickly.
• Delayed reaction to market-wide news: Stocks that react slowly to common factors catch up in the following days, creating the appearance of reversal in individual returns.

By excluding month t-1 from the formation window, the strategy avoids contaminating the genuine medium-term momentum signal with this short-term reversal noise. The standard implementation ranks stocks by their cumulative return from month t-12 to month t-2, then holds the portfolio for one month before rebalancing.

Jegadeesh and Titman returned to the topic in their 2001 follow-up paper, “Profitability of Momentum Strategies: An Evaluation of Alternative Explanations,” published in the Journal of Finance, Vol. 56, No. 2, pp. 699–720. Using an extended sample through the 1990s, they confirmed that momentum investing profits persisted in the post-publication period, effectively ruling out data-mining as the primary explanation.

4. Why Momentum Works: Behavioral and Risk-Based Explanations

The persistence of momentum returns has generated a rich debate about underlying mechanisms. Explanations fall into two broad camps: behavioral and risk-based.

Behavioral Explanations

Investor underreaction to news. Harrison Hong and Jeremy Stein's 1999 model, published as “A Unified Theory of Underreaction, Momentum Trading, and Overreaction in Asset Markets” in the Journal of Finance, Vol. 54, No. 6, pp. 2143–2184, proposes that information diffuses gradually across the investor population. “Newswatchers” trade on fundamental information but ignore past prices, while “momentum traders” extrapolate from past prices but ignore fundamentals. The slow diffusion of information among newswatchers creates initial underreaction, generating the momentum effect. Eventually, momentum traders pile in and push prices past fundamental value, creating the long-horizon reversal that Jegadeesh and Titman documented.

The disposition effect. Andrea Frazzini's 2006 paper, “The Disposition Effect and Underreaction to News,” published in the Journal of Finance, Vol. 61, No. 4, pp. 2017–2046, identified a concrete mechanism for underreaction. The disposition effect — investors' tendency to sell winners too early and hold losers too long — was first documented by Shefrin and Statman (1985) and confirmed with trading data by Odean (1998). Frazzini showed that stocks where mutual fund holders have large unrealized capital gains (and therefore are more likely to sell prematurely after good news) exhibit stronger momentum. The premature selling creates selling pressure that prevents the price from fully adjusting to positive information, generating a predictable drift.

Confirmation bias and limited attention. Investors anchor on their existing beliefs and underweight disconfirming evidence. When a stock has been rising, investors who already own it seek confirming information, while those who don't own it fail to pay attention to accumulating positive signals. This creates gradual price discovery rather than the instantaneous adjustment predicted by efficient markets theory.

Risk-Based Explanations

Risk-based explanations argue that momentum returns compensate investors for bearing some form of time-varying risk. However, these explanations have faced significant challenges.

Fama and French (1996), in “Multifactor Explanations of Asset Pricing Anomalies,” published in the Journal of Finance, Vol. 51, No. 1, pp. 55–84, tested whether their three-factor model (market, size, value) could explain momentum returns. It could not. They explicitly acknowledged momentum as the “main embarrassment” of their model and called it the “premier anomaly” that their framework left unexplained. This was a remarkable admission from the architects of the most widely used factor model in finance.

Later risk-based models have had more success. Johnson (2002) proposed a model where expected growth rates are stochastic and persistent, generating momentum-like patterns in returns. Liu and Zhang (2008) showed that an investment-based model with growth rate risk could explain about half of momentum profits. But no purely risk-based model has provided a fully satisfactory explanation, and the behavioral evidence remains compelling.

5. The Risks: Momentum Crashes

If momentum were a free lunch, it would have been arbitraged away long ago. It is not. The strategy carries severe left-tail risk in the form of momentum crashes — sudden, violent reversals where past losers dramatically outperform past winners.

The definitive study of this phenomenon is Kent Daniel and Tobias Moskowitz's 2016 paper, “Momentum Crashes,” published in the Journal of Financial Economics, Vol. 122, No. 2, pp. 221–247. Daniel and Moskowitz documented that momentum strategies have experienced infrequent but catastrophic drawdowns, typically occurring when the market rebounds sharply after a prolonged decline.

The mechanism is straightforward. During bear markets, the momentum portfolio's short leg consists of the biggest losers — typically high-beta, financially distressed stocks. The long leg consists of defensive, low-beta stocks that held up relatively well. When the market suddenly reverses (as it did in March 2009), the distressed stocks in the short leg explode upward while the defensive longs lag. The result is a devastating loss.

Historical Momentum Crashes

The 2009 crash was particularly devastating. Daniel and Moskowitz documented that from March to May 2009, the long-short momentum portfolio in US stocks lost approximately 40%. The short leg (past losers, dominated by financials like Citigroup and Bank of America) surged over 100% as the market bottomed, while the long leg of defensive winners barely participated in the rally.

Daniel and Moskowitz showed that momentum crash risk is partially predictable. Crashes tend to occur when (1) the market has recently experienced high volatility, (2) the market has recently declined sharply, and (3) the spread in betas between the winner and loser portfolios is wide. They proposed a dynamic momentum strategy that scales exposure inversely with forecasted volatility, which significantly reduces crash risk while preserving most of the strategy's long-run returns. This insight has been widely adopted by quantitative firms.

The implication for practitioners is clear: applying leverage to a static momentum strategy is extraordinarily dangerous. Because momentum crashes are correlated with market stress — precisely when margin calls are most likely — leveraged momentum traders face a lethal combination of drawdown and forced liquidation. Survival through a 2009-style event on leveraged capital is improbable.

6. Global Evidence: Momentum Everywhere

The momentum effect is not a quirk of US equity markets. It has been documented across virtually every liquid asset class and geography tested.

K. Geert Rouwenhorst's 1998 paper, “International Momentum Strategies,” published in the Journal of Finance, Vol. 53, No. 1, pp. 267–284, tested momentum in twelve European markets (Austria, Belgium, Denmark, France, Germany, Italy, Netherlands, Norway, Spain, Sweden, Switzerland, and the United Kingdom) over the period 1980 to 1995. He found an internationally diversified momentum strategy that earned approximately 1% per month, mirroring the US evidence. Momentum was present in every individual country sample, though the magnitude varied. Importantly, international momentum was not explained by country-specific factors, size, or book-to-market.

The broadest evidence comes from Clifford Asness, Tobias Moskowitz, and Lasse Heje Pedersen's 2013 paper, “Value and Momentum Everywhere,” published in the Journal of Finance, Vol. 68, No. 3, pp. 929–985. This landmark study tested both value and momentum strategies across eight diverse markets and asset classes:

• Individual stocks in the US, UK, Continental Europe, and Japan
• Equity index futures across developed markets
• Government bond futures
• Currency forwards
• Commodity futures

They found that momentum generated positive returns in every single asset class tested. The cross-asset universality of momentum is powerful evidence against data-mining explanations and suggests that the underlying behavioral mechanisms — underreaction, disposition effects, herding — are fundamental features of how humans process information and make financial decisions, regardless of the specific asset being traded.

Asness, Moskowitz, and Pedersen also documented a striking pattern: value and momentum are negatively correlated within asset classes but positively correlated across asset classes. This suggests that a common set of factors drives both anomalies, and that a combined value-momentum portfolio offers significant diversification benefits.

7. Post-Publication Performance and Decay

A persistent concern with any published anomaly is that once the academic paper is widely known, traders will exploit the strategy and compete away its profits. R. David McLean and Jeffrey Pontiff addressed this question directly in their 2016 paper, “Does Academic Research Destroy Stock Return Predictability?” published in the Journal of Finance, Vol. 71, No. 1, pp. 5–32.

McLean and Pontiff studied 97 variables shown to predict cross-sectional stock returns in published academic papers. They found that portfolio returns based on these predictors declined by an average of 26% out-of-sample (after the original sample period ended but before the paper was published) and by 58% post-publication (after the paper appeared in an academic journal). The out-of-sample decay suggests some original data-mining or structural changes, while the additional post-publication decay is consistent with informed trading eroding the anomaly.

For momentum specifically, the picture is more nuanced. While absolute returns have declined from the roughly 1% per month documented in the original 1993 paper, the strategy has not been arbitraged away. This is likely because momentum's severe crash risk creates a natural limit to arbitrage — rational traders are unwilling to fully exploit the strategy because doing so exposes them to catastrophic losses during reversals.

Recent evidence suggests that momentum remains most robust in mid-cap stocks, where liquidity is sufficient for institutional implementation but where information processing is slower than in mega-caps. In the largest, most heavily analyzed stocks, momentum profits have declined substantially as algorithmic traders and quantitative funds have increased their market share.

8. Practical Implementation Considerations

Moving from academic backtests to a live momentum trading strategy requires careful attention to several practical issues that can substantially erode or eliminate paper profits.

Transaction Costs

Momentum is a relatively high-turnover strategy. Monthly rebalancing requires selling all positions that have dropped out of the top decile and buying new entrants. Robert Korajczyk and Ronnie Sadka, in their 2004 paper “Are Momentum Profits Robust to Trading Costs?” published in the Journal of Finance, Vol. 59, No. 3, pp. 1039–1082, showed that the price impact of momentum trades is substantial, especially for smaller and less liquid stocks. They estimated that momentum profits vanish entirely for large institutional orders in illiquid stocks.

Their analysis suggested that a momentum fund could be profitable up to a capacity of approximately $2 to $5 billion in the US market (in 2004 dollars), depending on implementation details. Beyond that scale, the fund's own trading would erode its returns. This is a critical finding — momentum is a capacity-constrained strategy.

Portfolio Construction Best Practices

Several implementation choices significantly affect realized momentum returns:

• Value-weighted, not equal-weighted portfolios. Equal-weighted momentum portfolios tilt heavily toward micro-cap stocks where momentum is strongest on paper but where transaction costs, short-sale constraints, and illiquidity make implementation impractical. Value-weighting produces more conservative but more realistic returns.
• NYSE breakpoints. Hou, Xue, and Zhang (2020), in “Replicating Anomalies,” published in the Review of Financial Studies, Vol. 33, No. 5, pp. 2019–2133, emphasized that using NYSE breakpoints (rather than breakpoints based on all NYSE, AMEX, and NASDAQ stocks) prevents the portfolios from being dominated by tiny, illiquid NASDAQ stocks. This is now standard practice in factor construction.
• Excluding micro-caps. Applying a minimum market capitalization filter (e.g., above the 20th percentile of NYSE market cap) dramatically improves the implementability of the strategy.
• Overlapping portfolios. Rather than concentrating all rebalancing on a single day, using overlapping cohorts (e.g., forming one-third of the portfolio each month, each held for three months) smooths turnover and reduces the market impact of trading.

Combining Momentum with Other Factors

One of the most robust findings in the factor investing literature is that momentum combines well with value. Since the two factors are negatively correlated (cheap stocks tend to be recent losers, expensive stocks tend to be recent winners), a portfolio that combines both captures the benefits of each while significantly reducing overall volatility and drawdown risk. Asness, Moskowitz, and Pedersen (2013) demonstrated that this negative correlation holds across asset classes and geographies.

Additionally, combining momentum with quality factors (profitability, low leverage, stable earnings) tends to reduce the severity of momentum crashes, since the quality filter screens out many of the financially distressed stocks that drive the crash dynamics documented by Daniel and Moskowitz.

9. Scanning for Momentum with Alpha Suite

Alpha Suite's JT Momentum scanner ranks 130+ liquid US equities by their 12-2 month cumulative returns, implementing the canonical Jegadeesh-Titman specification. The system computes momentum scores daily, applies a minimum liquidity filter, and overlays the rankings with insider trading signals from SEC Form 4 filings.

When corporate insiders are buying shares in stocks that also rank highly on momentum — or selling shares in stocks with collapsing momentum — the convergence of these two independent signals can provide higher-conviction trade ideas than either signal alone. This multi-factor approach aligns with the academic evidence that combining uncorrelated return predictors improves portfolio Sharpe ratios.

The platform also implements the dynamic volatility scaling approach suggested by Daniel and Moskowitz (2016), reducing momentum exposure during periods of elevated market volatility to mitigate the risk of momentum crashes.

References

1. Asness, C. S., Moskowitz, T. J. & Pedersen, L. H. (2013). “Value and Momentum Everywhere.” The Journal of Finance, 68(3), 929–985.
2. Daniel, K. & Moskowitz, T. J. (2016). “Momentum Crashes.” Journal of Financial Economics, 122(2), 221–247.
3. Fama, E. F. & French, K. R. (1996). “Multifactor Explanations of Asset Pricing Anomalies.” The Journal of Finance, 51(1), 55–84.
4. Frazzini, A. (2006). “The Disposition Effect and Underreaction to News.” The Journal of Finance, 61(4), 2017–2046.
5. Hong, H. & Stein, J. C. (1999). “A Unified Theory of Underreaction, Momentum Trading, and Overreaction in Asset Markets.” The Journal of Finance, 54(6), 2143–2184.
6. Hou, K., Xue, C. & Zhang, L. (2020). “Replicating Anomalies.” The Review of Financial Studies, 33(5), 2019–2133.
7. Jegadeesh, N. & Titman, S. (1993). “Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency.” The Journal of Finance, 48(1), 65–91.
8. Jegadeesh, N. & Titman, S. (2001). “Profitability of Momentum Strategies: An Evaluation of Alternative Explanations.” The Journal of Finance, 56(2), 699–720.
9. Korajczyk, R. A. & Sadka, R. (2004). “Are Momentum Profits Robust to Trading Costs?” The Journal of Finance, 59(3), 1039–1082.
10. McLean, R. D. & Pontiff, J. (2016). “Does Academic Research Destroy Stock Return Predictability?” The Journal of Finance, 71(1), 5–32.
11. Rouwenhorst, K. G. (1998). “International Momentum Strategies.” The Journal of Finance, 53(1), 267–284.