Earnings Key in Momentum Outperformance

• High-momentum stocks, or those characterized by strong positive past price, high actual and forecasted earnings growth and positive market sentiment trends, tend to outperform low-momentum stocks.
• High-momentum stocks also tend to outperform the broader market over time and across geographies.
• Our research suggests the momentum factor can play a useful role in the stock selection process and, given its low correlation to other common factors, provide diversification benefits in portfolio construction.

Recent broad swings in performance highlight the role market momentum can play in driving stock performance. A stock’s momentum, amplified due to information inefficiencies and delays in information processing, can result in an overreaction, underreaction or even delays in stock performance relative to company-specific events such as earnings. Additionally, bullish investor sentiment can be a powerful driver of stock prices, as behavioral biases behind herd mentality, trend chasing and anchoring can result in higher future returns.

To illustrate the effects of momentum, we theorize that stocks with a high momentum will continue their strong performance and generate a higher return than stocks with a low momentum. Incorporating established academic research¹, we defined momentum as including both price and earnings momentum; we also looked at data from trading activities and the option market to gauge investor sentiment. We tested a number of subfactors that measured momentum and, on the basis of our economic theory and back-tested results, used six to identify and characterize stocks as having high momentum (Exhibit 1).

Exhibit 1: The Subfactors that Form the Momentum Factor

We also tested additional subfactors, including trailing one-month, three-month and six-month price return, information discreteness of stock returns, the three-day price impact of earnings surprises, inter-industry momentum, and the effect of turnover on short-term price momentum. However, we found that these subfactors had weaker economic rationale and performance compared to the six subfactors listed, and thus we did not include them in the overall momentum factor.

Thus, a stock with high positive momentum would be characterized by strong positive past price trends, high actual and forecasted earnings growth, low short interest and higher implied volatility spread of call options versus put options.

We began with a universe of all common stocks, REITs and MLPs listed on U.S. exchanges while excluding ADRs. We then filtered out stocks whose trailing three-month average daily volume (ADV) was below the 25th percentile and whose market cap was below the 20th percentile of the Russell 3000 Index. This translated to a minimum tradeable ADV of about $4.11 million and minimum market cap of about $549 million as of June 2022. By doing so, we tested our factor using  only stocks in which an active institutional equity manager portfolio could realistically have taken meaningful positions.

We generated scores for each of the six subfactors by calculating the standardized rank of a stock’s subfactor relative to the universe. For example, a stock that had a trailing 12-month return within the top (bottom) 20% of the investable universe would receive a quintile rank of 1 (5) for the 12-month price momentum subfactor. The overall momentum score was then computed as an average of the standardized ranks of its six underlying subfactors relative to the universe within each size bucket. The top quintile of momentum scores was classified as high momentum, while the bottom quintile was classified as low momentum.

We found that between January 1996 (when sufficient data for our purposes became available) to June 2022, portfolios consisting of high-momentum stocks consistently outperformed those consisting of low-momentum stocks. Exhibit 2 shows the excess return of being long the high-momentum stocks and short the low-momentum stocks in various size buckets. Exhibit 3 displays the cumulative all cap long-short value-weighted returns of the momentum factor. 

Exhibit 2: Annualized Excess Returns of the Long-Short Momentum Factor

Exhibit 3: Cumulative Returns of the Long-Short Momentum Factor

Exhibits 4 and 5 display the average annualized and cumulative (respectively) value-weighted forward returns of a long-only high-momentum portfolio comprising stocks ranked in the top quintile of momentum scores against its benchmark.

We can make a few observations. For example, the performance of the overall momentum factor was robust across different size cohorts. Furthermore, the momentum factor had stronger performance than most of its subfactors (Exhibit 6).

By aggregating the subfactors, the momentum factor improves the performance of the individual subfactors, even though some individual subfactors might perform better in subsamples of the data.

We also regressed the monthly returns of the momentum factor and its subfactors against the Fama-French three-factor (FF3) and five-factor (FF5) models to test the momentum factor return excluding the impact of other common factors, including the market, size, value/growth, profitability and investment. Exhibit 7 displays the annualized regression coefficients (also known as regression alphas) and t-stats for the back tests. Although the subfactors had variable alphas, the momentum factor had statistically significant positive alphas with respect to both Fama-French factor models.

The momentum factor has an annualized alpha of 16.78% and a significant t-stat of 5.01 with respect to the FF3 model, and an annualized alpha of 14.26% and a significant t-stat of 4.14 with respect to the FF5 model. Thus, the momentum factor consistently delivered positive risk-adjusted returns and is a robust factor distinct from other common factors at the 5% level of significance (at which level of testing a t-stat of 1.96+ indicates significance).
 

Exhibit 4: Annualized Returns of the Long-Only Quintile 1 High-Momentum Portfolio and Russell 3000

Exhibit 5: Cumulative Returns of the Long-Short Momentum Factor

Exhibit 6: Annualized Returns of the Momentum Factor and its Subfactors

The momentum factor had statistically significant negative coefficients with the market (MKT) and high book value to low book value (HML), also known as value, factors and provides diversification benefits (Exhibits 8 and 9). The coefficients with the small minus big (SMB), robust minus weak (RMW) and conservative minus aggressive (CMA) factors are positive but not statistically significant.

The correlation matrix between the momentum factor and its subfactors indicates that the subfactors provide distinct signals that characterize high-momentum stocks (Exhibit 10). The correlations of the price momentum subfactors (Price12M and Price3M) and earnings momentum subfactors (SUE and REV6) were moderately high. This is to be expected, as high earnings growth may drive high returns. The investor interest (SI and OPTVOL) subfactors had relatively low correlation to other subfactors.

We also back tested the equal-weighted, industry neutral, and sector neutral forward returns of momentum and its subfactors, and we find that the conclusions are similar to the value-weighted forward returns discussed above. We also constructed a global version of the momentum factor using a universe of both U.S. and international stocks and found that the global momentum factor was also robust in both the base specification and country neutralized specification. Thus, the momentum factor is robust in both the U.S. and global universe, using different weighting schemes, and after removing sector-, industry-, and country-specific effects.

Exhibit 7: Annualized Regression Alphas of the Momentum Factor and Subfactors Against Fama-French Factors (FFF)

Exhibit 8: Regression Coefficients of the Momentum Factor Against Fama-French Factors

Exhibit 9: Regression t-stats of the Momentum Factor Against Fama-French Factors

Exhibit 10: Correlation Matrix of the Momentum Factor and its Subfactors

The momentum factor exhibits strong and robust performance, and our back tests provide support for the robustness and distinctiveness of the factor. We found that across market cap groups, high-momentum stocks identified in Quintile 1 tend to generate the highest annualized average returns, while low-momentum stocks identified in Quintile 5 tend to generate the lowest returns (Exhibit 11). From January 1996 through June 2022, the all cap annualized return of the Quintile 1 high-momentum portfolio was 12.0% and was comparable to the long-short factor return of 12.9%. This indicates the momentum factor was not solely driven by the short Quintile 5 low-momentum portfolio. Additionally, the annualized standard deviation of all cap returns was lower among high-momentum stocks (18.2%) than low-momentum stocks (24.9%).

The performance of the momentum factor remains robust in global markets as well. Expanding our investment universe to include the S&P Global BMI Index and FTSE Global Index, we found that the momentum factor generated significant positive alphas with respect to the Fama-French models, illustrating that the global momentum factor continued to deliver positive risk-adjusted returns and is a robust factor distinct from other common factors. Additionally, to further test the robustness of the global momentum factor, we neutralized country-specific effects to test whether the factor’s robustness was geographically dependent and found that the momentum factor continued to persist in global equity markets. 

Exhibit 11: Average Annualized Returns in the Momentum Factor Quintiles

Exhibit 12: Average Annualized Returns in the Momentum Factor in Various Time Periods

We also examined the average annualized return of the momentum factor in three non-overlapping nine-year time periods. We found that high-momentum stocks (Q1) consistently outperformed low-momentum stocks (Q5), and the trend of higher ranked stocks outperforming lower ranked stocks persisted through all five quintiles. Thus, the momentum factor is also robust in different time periods. 

We believe the momentum factor can be a useful addition to both the stock selection and risk management processes. Our back tests provide compelling evidence that high-momentum portfolios outperformed low-momentum portfolios in aggregate, and the factor was also robust in different model specifications. The 12-month price momentum, three-month price momentum, standardized unexpected earnings, changes in earnings forecast, short interest and option volatility spread subfactors help to identify the key characteristics of high-momentum stocks, making the factor a useful screening tool for fundamental stock pickers. The statistically significant positive alphas of the long-short factor and outperformance of the long-only high-momentum portfolio against the benchmark further indicate that the factor provides a potential source of excess returns in portfolio construction. Finally, the momentum factor also provides diversification benefits against other common factors.

About the Authors

Farhan Mustafa, CFA

Head of Investment Risk Management, Head of Quantitative Research

19 Years experience            19 Years at ClearBridge

Yaoqin Li, CFA

Quantitative Research Associate

7 Years experience            4 Years at ClearBridge