zondag 14 mei 2017

Quants Are Looking At Increasingly Stranger Things For A Trading Edge

Bank of America released its annual "Institutional Factor Survey", i.e., questionnaire targeting market quants (but not only), which seeks to assess which factors and attributes are most important in the current environment to clients in selecting securities. As it introduces the 37 page report, "even if you don’t wear a pocket protector this report is relevant" as the rise in “smart beta” assets at close to a 30% annualized run rate during this bull market and what feels like a seismic shift in assets from fundamental to systematic strategies - suggest factor investing matters to everyone, not just quants. She is right: as we predicted all the way back in 2009, in a world controlled by algos and where speed of trading matters more than depth of analysis, the quants have now taken over. This is shown by the next chart which confirms that some time in 2014-2015, fundamental investing became of secondary interest to the investing community a watershed event with huge implications and which has received little attention in the broader press...


*) As BofA's Savita Subramaian, who conducted the study notes, the results of our 26th annual client survey show which factors clients are using relative to history, and where we could see the next bubble. "Case in point: we saw a rise in the popularity of beta in our survey beginning in 2007, which presaged the remarkable growth in “Low Vol” ETFs and subsequent bubble-like valuations of low beta stocks"...
 

Furthermore, BofA's survey found that nearly two-thirds of modern investors use a mix of quantitative and fundamental signals. But is everyone using the same models? The survey also showed that quant models are growing more complex, the number of signals used by the avg. respondent is 17, more than double the factors cited in the early 90s and correlations between clients have generally decreased.
*) So before we get to the current survey, BofA recalls that last year it noted that Value factors were by far the most popular, and sure enough, Value overtook both Yield and Growth as the biggest “smart beta” ETF category by AUM, with 26% of assets today vs. 23% a year ago. This could continue: Value factors remain the most popular category in our 2017 survey, even more so than last year. Low P/E once again topped the list of the most popular factors used by clients (Chart 1), and continued inflows into Value strategies may provide the technical underpinnings for a longer Value cycle, which we expect for a host of other reasons, both macro and fundamental. In light of the above, there were few surprises when looking at the most popular existing factors used by the quant community. Nearly 80% of survey respondents use Forward P/E as factor in their investment processes, remaining the most popular factor for the 11th year running (and perhaps explaining why the market no longer makes sense when wave after wave of short squeezes of the most overvalued stocks crushes even the most determined fundamental investors). While cash-flow based valuation measures were more popular pre-crisis, Forward P/E has topped the list every year since the crisis. EV/EBITDA also remained popular (second-most cited factor), while Net Debt/EBITDA continued to climb on the list (more on this below)...


So where do we stand now? Well, given that alpha signals tend to be exploited quickly and arbitraged away, quants are increasingly focused on signals which as recently as a decade ago would make little to no sense for virtually any investors: among the new factors are real-time data feeds, artificial intelligence (AI), big data, etc. But we are in the early days: these were among the least-cited factors in our survey, particularly AI (used by just 7% of respondents). Web-scraping and machine learning were the most popular of these tools. But popularity is likely to grow: text-based algorithms saw the biggest increase in usage of any factor in our survey vs. a year ago. In our view, this is where the quant landscape is growing increasingly crowded. With that, on to the answer of the main question: how are Quants investing today, and which factors are the most popular today, in other words, what has grown most in popularity of the past year. Below BofA ranks factors (excluding new additions to our 2017 survey) based on the difference in popularity vs. 2016 (Chart 17). The factor which saw the biggest decline in popularity vs. a year ago was Share Repurchase, where just 42% of respondents cited using this factor, vs. 56% last year and over 60% in 2015. A more selective focus on buybacks makes sense, where companies doing the largest buybacks are no longer generating alpha, as a result, the Share Repurchase factor has been underperforming since late 2014. The median S&P 500 stock is trading at its highest levels since the Tech Bubble, and we’ve found that share buybacks tend to be most rewarded when stocks are inexpensive. Buybacks have also been slowing as investors have increasingly agitated for capex, and companies have also finally begun to see improvement in top-line growth, resulting in less of a need to prop up EPS via buybacks...


*) Some more details: Models more complex, new tools still early in uptake: Since 2007, investors have increasingly used a broader array of signals in their models in a quest for alpha, the average number of factors used reached a record high of 20 in 2015, before dipping down a bit to 16 last year. This year that number ticked up again, with respondents using an average of 17 factors, well above the historic average. Multi-factor models are more complex/diverse today than in the 1990s, when investors used an average of just 7-8 factors. And correlations between clients have generally decreased in recent years.
#To summarize, quants are increasingly focused on new factors, real-time data feeds, artificial intelligence, big data, machine learning, etc, as new alpha signals tend to be exploited quickly and arbitraged away. These techniques appear to still be early in their uptake, as they were among the least-cited factors used by 2017 survey respondents, particularly Artificial Intelligence, which was used by just 7% of respondents. Web-scraping and machine learning was the most popular quantitative tools, used by 19% and 13% of this year’s respondents, respectively. But we suspect these tools will continue to increase in popularity, in fact, text-based algorithms saw the biggest increase in usage of any factor between 2016 and 2017.
# Finally, BofA asks an interesting question: what are "the people with the pocket protectors" afraid of, or said otherwise, what do quants see as the biggest perceived threats to their model. The answer: short-term focus and crowded trades. 53% of investors cited the short-term focus of the investment community as the biggest risk to their investment processes, up from 50% last year (when it was also the most-cited threat)...


Ironically, BofA concludes that fundamental investing may be the most contrarian opportunity right now amid the rise of short-term investment strategies, particularly as fundamental strategies tend to win over the long haul, for example, valuations explain over 80% of S&P 500 returns over a 10-year time horizon. Crowding was the next most cited response (by 42% of investors), also the case in 2016. This also is an opportunity to use positioning as an input into the investment process, given that the most crowded stocks have underperformed the most neglected stocks for the last several years (except for YTD). Continued flows from active to passive vehicles suggest positioning will likely continue to matter....