6. Ratio of Up vs. Down FY1 EPS Estimate Revisions

Description of Factor:

This factor considers the analyst revisions to 1-year forward EPS estimates over the past twelve months. Specifically, this factor is calculated by: (number of upward revisions – number of downward revisions) divided by total number of estimates. Note that this factor does not consider the size of the revisions, just the directional change of the revised estimates. The rationale behind this variable is the expectation that stocks with more upward revisions than downward revisions will exhibit earnings momentum and earnings surprise characteristics. We surmise that stocks which contain a high value for this variable will outperform stocks with low or negative values for this variable.

Analysis

Equal Weighted: Looking at the annualized average quintile return graph below, a step distribution is exhibited. However, there is very little spread between the returns of quintiles 3-5. The spread between quintile 1 and 5 is quite small at 6.46%. Quintile 1 has the highest annualized average return of 23.43% with a standard deviation of 16.80%, while quintile 2 exhibits the lowest annualized average return of 16.97% and standard deviation of 13.97%. The average excess return of quintile 1 is 3.76% and (2.06)% for quintile 5. The beta for quintile 1 is 1.10 with an R² of 73%, while beta for quintile 5 is 0.91 with an R² of 72%.

The annual returns for a long-short strategy based on this factor are almost always positive except for 2001 and 2002. However, the losses in 2001 and 2002 were quite small at (0.38%) and (3.07%) respectively. This result is very important from a capital preservation perspective. It follows logically that the small expected spread of 6.46% stated above is too minute to produce adequate long-term capital appreciation with investments in portfolios with equity betas above 1.0 in each quintile. A notable achievement of this strategy is a significant positive return in both 1999 and 2003 (16.44% and 25.31% respectively). This is notable because though the two markets were quite dissimilar, quintile 1 and 5 achieved the desired top-performing and bottom-performing positions respectively in both years. However, results in recent years are mixed as quintile 1 narrowly beat quintile 5 in 2000 and lost in 2001 and 2002, as discussed above. Perhaps the most intriguing aspect of this strategy is the extremely low turnover ratios. Quintile 1 turns over only 3.63% of its portfolio companies each month while quintile 5 enjoys a 4.39% rate.

Value Weighted: A value weighted portfolio does not exhibit a downward step distribution, with the annualized annual returns of quintile 4 eclipsing those of quintiles 2 and 3. The spread between quintile 1 and quintile 5 is only 4.19%. Quintile 1 has the highest annualized average return at 21.21% and standard deviation of 14.16%. Quintile 5 has the second-lowest average annual return of 17.02% and lowest standard deviation of 12.54%. The average excess return of quintile 1 is 1.72% and (2.06)% for quintile 5. The beta for quintile 1 is 1.02, while beta for quintile 5 is 0.90.

A long-short strategy based on this factor using value weights would produce a negative return 7 times out of 16 sampled years and 2 out of the last 4 years. Only three times does quintile 1 achieve the highest return and quintile 5 achieve the lowest return in the same year. Quintile 1 attains the highest return 5 times and the lowest return once. Quintile 5 achieves the lowest return six times and the highest return twice.

Conclusion:

The equal weighted strategy based on this variable performs much better than a value weighted strategy. This philosophy seems practical to implement due to its low turnover ratios and limited downside. We are pleased with its effectiveness in two up-markets (1999 and 2003) and its relatively safe returns in three turbulent markets (2000, 2001, and 2002). While there are some extremely positive returns, most notably in 2003, the majority of the returns have been far too small to adequately satiate investors who are seeking equity market returns. This variable clearly exhibits a track record of solid results. Though we believe it is very promising when used with an additional variable, we do not recommend a trading strategy based on it alone.