We continue last week’s introduction to the Reversal factors found in the Barra US Total Market Equity Trading Model and discuss how these factors can be useful in helping guide trading. Although the risk model has 3 flavors of reversion, we will focus on the two most relevant to fundamentally oriented and longer horizon investors: Short-Term Reversal (“STR”) and Long-Term Reversal (“LTR”). For a refresher on these factors and their characteristics, please revisit last week’s post.Last week, we observed the long-term trends in these factor returns: Since 2008, Short-Term Reversal has outperformed the market by +3.5%/annum and Long-Term Reversal has underperformed the market by -0.7%/annum. Accordingly, having a positive exposure to Short-Term Reversal names and a negative exposure to Long-Term Reversal names will help portfolio returns and having a negative exposure to Short-Term Reversal and a positive exposure to Long-Term Reversal will hurt portfolio returns.

Thus, our hypothesis is simple: your stock is a candidate for mean reversion if it contains both positive STR and negative LTR. Below we test this hypothesis and highlight our findings.

Designing Mean Reversion Portfolios

We start by building a basic filter to define our local universe for the most liquid and tradeable names in the US, roughly equivalent to the Russell 3000. The cap-weighted universe has an average of 2,814 names and roughly a 2.5% tracking error to the Russell 3000, until the tracking error exploded during the COVID crisis. We then further filter this investible universe down to two groups depending on the Short-Term Reversal and Long-Term Reversal exposures:

'pos” refers to stocks with favorable exposures: positive to STR and negative to LTR.

“neg” refers to assets with unfavorable exposures: negative to STR and positive to LTR.

Once we have these stocks identified, we then build two portfolio variants: one market-cap weighted and one factor weighted. Factor-weighting assigns an asset’s weight dependent upon it’s factor exposures rather than its market capitalization. In our case, we weighted our '“pos” and “neg” portfolios by simply summing the absolute value of each exposure. Assets with more extreme exposures accordingly have a higher weight in the portfolio, a la Research Affiliates. We build these 4 portfolios on a monthly basis, starting on 5/31/2018.

Examining Mean Reversion Portfolios

Below we confirm that our portfolios have the desired exposures to LTR:

• Pos_mcap and Pos_factor are negatively exposed to LTR
• Neg_mcap and Neg_Factor are positively exposed to LTR

Below we confirm that our portfolios have the desired exposures to STR:

• Pos_mcap and Pos_factor are positively exposed to STR
• Neg_mcap and Neg_Factor are positively exposed to STR

This chart illustrates how quickly the Short-Term Reversal switches signs relative to the Long-Term Reversal factor. The Short-Term Reversal factor moves quickly, which induces turnover and transaction costs. As discussed last week, this variation of Short-Term Reversal “washes” out in the long term for long horizon investors.

How have these portfolios performed?

In both the “mcap” and “factor” weighting the POS portfolios outperform their respective NEG portfolios by ~8.5%/annum. This is obviously a very strong signal, and reinforces how quant flows have become a strong component of the market.

Concluding Observations

We run this experiment to illustrate the general market trend of how Reversal Factors behave. A Fundamental stock picker would never build a portfolio in this manner - it’s much more of a Quantitative Manager’s approach. But, we can at least highlight assets that have favorable trends and help managers decide buy and sell timing on names they have researched and either want to buy, or sell out of.

For Quants: It is worth acknowledging that the 50%/50% blend of both Short-Term Reversal and Long-Term Reversal may not be the best way to combine signals. Trimming extreme outliers, weighting the more stable Long-Term Reversal factor, and other approaches to building a factor weighted portfolio - especially when combined with factor-awareness would likely improve the factor weighted portfolio.

And never fear! Even if an asset you want to buy has a negative exposure to Long-Term Reversal and negative exposure to Short-Term Reversal, just wait ~ 20 trading days for the Short-Term Reversal exposure to swing directions!

US Market: 5/18/20 - 5/22/20

• The market ended flat on the Friday before the holiday weekend, after posting strong gains throughout the week as businesses in all 50 states started to reopen, and despite heightened US-China tensions.
• Investor optimism was buoyed by encouraging initial results of a COVID-19 vaccine by Moderna and comments from Dr. Fauci expressing confidence in a remedy for the disease by year end.
• To that end, the US is now planning a testing effort involving more than 100k volunteers and several of the most promising vaccine candidates.
• First-time unemployment filings last week came in at 2.44 million, in line with consensus estimates of 2.4 million. This brings the total to 38.6 million over the past 9 weeks, since the lockdown started.

Factor Update: Axioma US Equity Risk Model (AXUS4-MH)

• Growth remained the biggest winner, although only gained +0.11 standard deviations as it headed closer to becoming Extremely Overbought.
• Volatility ended the week a little higher than the previous peak of +2.79 on 5/12.
• After ticking up slightly last week, the decline in Size continued as it fell below three standard deviations below the mean.
• Market Sensitivity continued to decline from +2.86 SD above the mean on 5/6, still retaining an Extremely Overbought designation.
• After peaking at +2.02 SD above the mean on 5/5, Value continued to trend back towards the mean, and looks posed to exit Overbought territory.
• Profitability fell deeper into Extremely Oversold space after hitting a recent peak of +6.47 SD above the mean on 3/20.
• US Total Risk (using the Russell 3000 as proxy) saw a slight decline of 15bps.

 Factor Update: Axioma Worldwide Equity Risk Model (AXWW4-MH)

• Exchange Rate Sensitivity saw continued strength as it climbed higher after its recent plunge to -4.9 SD below the mean on 3/31.
• Size saw a mild bounce back after dipping below -3 SD below the mean.
• Growth ended the week flat after hitting a high of +2.62 SD above the mean on 5/20.
• Volatility has continued to slowly decline from a recent peak of +2.87 SD above the mean.
• Value trended down towards an exit from Overbought space after hitting a recent high of +1.85 SD above the mean on 5/7.
• Earnings Yield dug deeper into Oversold territory, falling -0.3 standard deviations.
• Profitability was again the biggest loser for the fourth week in a row as it approaches Extremely Oversold territory. This factor was +3.98 SD above the mean on 4/2.
• Global Risk (using the ACWI as proxy) decreased by 21bps.

Please let us know if you’d like to better understand and manage your exposure to any of the factors that we’ve discussed.

Regards,
Chris