Expanding Our Mean Reversion Palette

We continue our discussion from last week on the topic of mean reversion, this time through the lens of the MSCI Barra US Total Market Equity Trading Model.

Before we do that though, we wanted to revisit our analysis on Antero Midstream, our mean reversion pick from last week. As you can see below, Antero has seen a fairly strong selloff MTD, and has now fallen from a +3.5 exposure (top 1%) to a +0.44 (66th percentile) exposure in the Axioma Short-Term Momentum factor. Certainly a strong case of reversion!

MSCI Barra US Total Market Equity Trading Model

MSCI Barra's model has three “flavors” of mean reversion across different time horizons: 1-Day Reversal, Short-Term Reversal, and Long-Term Reversal. This week we’ll go over the basics of these mean reversion factors, and next week we will discuss how to potentially incorporate these factors as a signal to help guide trading.

‍1-Day Reversal: the return differences of stocks based on their reversal pattern over the last day. Assets with high positive exposure to 1-Day Reversal are ones that had a dismal yesterday. Assets with negative exposures to 1-Day Reversal are yesterday’s winners.

Below is the long term factor return for 1-Day Reversal: we see that this is factor that trends upwards from a long-term perspective:

Short-Term Reversal: the return differences of stocks based on their reversal pattern over the previous month. This factor is the most similar to Axioma’s definition of “Short-Term Momentum” (STM), as discussed in last week’s blog post. Note that in this case assets with the highest exposure to Short-Term Reversal will be assets that have performed the worst over the previous month. This is the equal and opposite definition of Axioma’s STM, and accordingly we see the Short-Term Reversal factor has a long trending upward positive factor return:

Long-Term Reversal: the return differences of stocks based on their reversal pattern over the previous 2 years. This factor is unique from the other reversal factors, as it has two components: Long-term relative strength and Long-term historical alpha. As most of your know, Relative Strength (or RSI) has been used by technical practitioners for decades. On the other hand, historical alpha is interesting because it measures assets that that have dramatically outperformed their peers over the recent 2 years even when removing the market, sector, and all style factors from the return.

Because assets that have outperformed the most will have a positive exposure, the consistently negative factor return will add a drag to performance. We can see that this factor is substantially weaker than the 1-day and short-term reversal, where quantitative strategies tend to be more active.

Preview for next week - how to combine these signals?

MSCI Barra’s three reversion factors are built to be independent of one-another. Thus, the returns you see above are additive and ultimately can be used to create a combined signal that can increase the confidence over detecting the potential mean reversion in an individual security or a basket of securities.

We can also see that the strongest factor is the 1-day reversal (~159% up since 1/1/2008), followed by the short-term reversal (~90% up since 1/1/2008), and the long-term reversal (~12.58% down since 1/1/2008). Thus, we would expect that when the short-term and 1-day reversal line up, the likelihood for reversion will become substantially stronger.

US & Global Market Summary

US Market: 5/11/20 - 5/15/20

• The market ended the week lower as investors digested disappointing retail news and another massive round of jobless claims, slightly tempered by the prospect of another round of stimulus.
• Jobless claims saw another 3 million Americans filing for benefits, topping 36 million since the beginning of the pandemic.
• The Commerce Department reported that US retail sales in April fell 16.4% vs. the consensus expectation of 12.5%.
• The Fed reported that industrial production fell 11.2% in April, the largest gap down in 101 years of data. This number still came in better than the -12.5% consensus forecast.
• Sentiment was buoyed at the end of trading Friday after Congress approved a $3 trillion stimulus package that will now have to make its way through the Senate.

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

• Growth was again the biggest winner as it saw a +0.69 standard deviation move, and is now at the threshold of being Extremely Overbought.
• Medium-Term Momentum saw some strength on a normalized basis, as it headed back towards the mean.
• The free-fall in Size appears to have abated after bottoming at -2.89 SD below the mean on 5/13.
• Volatility ended the week perfectly flat after peaking at +2.79 on 5/12, suggesting a potential reversion in the near term.
• Market Sensitivity appears to have peaked at +2.86 SD above the mean on 5/6 and has since started to move back down, still retaining an Extremely Overbought designation.
• After peaking at +2.02 SD above the mean on 5/5, Value has trended down from Extremely Overbought and now sits at 1.5 SD above the mean.
• Profitability continued to plummet from its recent peak of +6.47 SD above the mean on 3/20, and is now on the verge of becoming Extremely Oversold.
• US Total Risk (using the Russell 3000 as proxy) saw a slight decline of 30bps.

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

• Exchange Rate Sensitivity continued to see strength as it crossed into positive Neutral territory after another strong upward move. We discussed this factor in detail after it had plummeted to -4.9 SD below the mean on 3/31 (Part 1 and Part 2), and it’s interesting to consider what the impact of the next round of stimulus might have on this factor.
• Growth entered Extremely Overbought space and continued to climb higher, sitting at +2.53 SD above the mean.
• The decline in Size decelerated, although it is now below -3 SD below the mean.
• Volatility has started to come down from the recent peak of +2.87 SD above the mean.
• Earnings Yield fell into Oversold territory, although the pace of decline was slower than last week’s -0.85 SD move.
• Profitability was the biggest loser for the third week in a row and is now an Oversold factor at -1.23 SD below the mean. Recall that this factor was +3.98 SD above the mean on 4/2.
• Global Risk (using the ACWI as proxy) decreased by 34bps, in line with US risk.

Regards,
Chris