Continue active refreshing of this index's data?

Continue active refreshing of this index's data?

Blog Posts — June 11, 2019

Factor Performance Continued to Fall Short in Q2—and Constraints Only Made Things Worse

by Melissa R. Brown, CFA

Momentum’s recovery in Q2 was a welcome event for many factor-based investors. Unfortunately, that recovery did not extend to other commonly used factors and probably wasn’t enough to pull quant managers out of the hole. Making matters worse, implementation constraints likely hurt performance. Axioma’s monthly US Large-Cap Multi Factor Sample Factor portfolio (which constrains exposures to industries and factors not included in the alpha[1] input for the portfolio) lagged the market by 1.2% in April and May, with IR of -2.3. That brought the year-to-date return and IR to -3.9% and -3.7, respectively. Our “Alternative Attribution” portfolios helped shed some light on this underperformance.

Exhibit 1. Multi Factor Sample Portfolio Performance

Source: FTSE Russell

Axioma’s portfolios are constructed to calculate the impact of common portfolio manager constraints and the results are meant to be compared with the factor-mimicking portfolio (FMP). The FMP comprises a broad universe of stocks, is rebalanced daily, is long-short, and has unit exposure to the factor and no other exposures. For each of a few common factors, we provide a few alternatives: long-short rebalanced monthly, long-short rebalanced daily using only the Russell 1000 as the universe, and long-only, where the portfolio cannot be short more than the benchmark weight. We also look at a couple of heuristic-based portfolios that only invest in the top quintile or go long the top quintile and short the bottom. These alternatives are meant to 1) better reflect the factor exposure a manager can actually achieve, and 2) facilitate comparison with some common factor definitions that are not based on “pure” portfolios. See our recent paper for more details.

April-May Results

After an initial four tough months 2019, Momentum’s FMP had in May one of its best months since at least 1999, as did the factor in most other regions (although our focus here is the US). Unfortunately, this didn’t particularly help long-only managers, as the long-only constraint took away more than 200 basis points of performance. In other words, the factor worked far better on the short side, so selling poor Momentum was fine, but buying good Momentum did not particularly help. Most other constraints were not nearly as harmful.

Despite the promising performance for Momentum, even a manager who could have benefitted from the ability to short poor Momentum was likely to have been hurt by other commonly used factors. Axioma’s Profitability factor has had negative returns in six of the past seven months. And May’s return fell into the bottom 5% of monthly returns since 1999. The slight silver lining for Profitability is that a long-only FMP produced a slightly positive return.

The last time Value had two consecutive months of positive return was in January 2018, and even those instances produced only slightly positive returns. May’s return fell just outside the bottom decile. But as in Momentum, the long-only FMP lagged the long-short version’s substantially.

While Earnings Yield produced acceptable returns from October through January, it, too, has seen its fortunes reverse, and May’s return was one of the most negative since at least 1999. Although other factors saw better performance for the Russell 1000 universe than in the broader Russell 3000, that was not the case for Earnings Yield, where large-cap-only investors likely felt even more pain.

The bottom line: the Multi Factor portfolio was hurt by its exposures to Earnings Yield, Profitability and Value, but was also not able to get the return it should have from Momentum and Value owing to its inability to short. The contribution from Earnings Yield was further dragged down by the large-cap universe restriction, but that was offset by the better large-cap performance from the other factors. Rebalancing monthly seemed to have minimal impact.

Exhibit 2: Alternative Attribution Portfolios

Note: Factor returns are scaled to a factor exposure of 1, to make them more comparable
Source: FTSE Russell, Axioma

[1] The portfolio tilts on Momentum, Value, Earnings Yield and Profitability, and away from Size and Volatility, with each factor equally weighted.