The Investment Assembly Line:
Alpha Discovery and the Illusion of Automation
Executive Summary
Despite its popularity in the modern market lexicon, there is no such thing as “alpha generation.” Investment managers and hedge funds do not produce alpha —as if it were the proverbial special sauce. Alpha is indirectly generated: It is a symptom of inefficiencies caused by market structure; inadvertently conflicting regulations and regulatory regimes; information distribution and cognitive disparities; and naturally occurring anomalies and behaviors among various types of competing market participants.
|
 |
|
Quantitative research is nothing more than a highly systematic approach to pattern recognition, and quantitative trading is simply the application of automation to the detection of these patterns. To translate this axiom into the overused concept of “alpha”—or risk-adjusted returns in excess of a benchmark—quantitative research is another term for alpha discovery, and quantitative trading—itself a euphemism for many other activities, such as algorithmic trading—is another term for alpha capture.
For many strategies and markets, automation is now a competitive necessity; but over time, more and more manually operated processes and instruments in the investment strategy workflow will fall prey to automation, generally for the greater good. The changes implemented roughly 30 years ago with a few pioneering physicists have now directly or indirectly infiltrated even the most conservative long-only asset managers.
One of the strongest indications of the maturity of this era of automation is the growth of industrial-grade tools, which were initially built internally, to manage the full workflow process of strategy development and production—known as the investment assembly line. Although comprehensive investment strategy platforms are normally the result of years of in-house development, the vendor community is quickly attempting to fill the growing demand for the rapid discovery, development, and deployment of automated strategies.
There are four core steps in the investment strategy process that are independent whether or not the strategy is automated, manual, or somewhere in between: Data management, Alpha discovery, Alpha capture and Feedback loop. Each phase has its own set of market data and analytics and includes major opportunities and challenges.
Although the level of automation across strategies remains quite disparate (compare distressed debt to statistical arbitrage, for example), the processes among those strategies that are particularly well-suited for automation are highly automated, with the exception of alpha discovery. Until software can be programmed to spot new patterns (rather than find known patterns in new datasets), the automation of the alpha discovery process is unlikely to change.
The TABB Group Vision Note on The Investment Assembly Line:
Alpha Discovery and the Automation Illusion
This 20-page report discusses the rise of quantitative strategies, the development of the tools to support those strategies and the application of those tools to a broader set of strategies. The report defines three high-level flavors of alpha, sub-categories within those flavors and the dynamics between automation and alpha decay. The report also outlines critical requirements for any comprehensive strategy development architecture, why pioneers built their own technology and will most likely continue, and how there is still a viable market for vendor-based solutions.