There’s been a lot of hype about the expected disruptive impact of technology on various industries, jobs and careers. Suddenly everybody in the asset management industry is wondering: Will computers take over?
We perceive the impact of technology as less disruptive and game-changing, and perhaps we do so because much of the mystery resides in the terminology used to describe technology, rather than in the application of the technology.
Just consider the reality versus the mystery surrounding the following concepts:
- Algorithm: A series of actions taken by software to accomplish various repetitive functions. These algorithms can be lengthy, complex and combined to cover a wide variety of possibilities. The fact remains that algorithms can only do what they are programmed to do. The programmer acts in conjunction with an expert and creates a set of decision trees and routines that replicate the typical decision-making process to analyse the specific situation. The algorithm will do this millions of times faster and cover much more material than a person can, but it can only make decisions within the framework of its programming.
- Machine learning: With much more data available than ever before, analyses can be performed which weren’t even considered previously. Patterns, causes and effects can be found where none were considered possible or even envisaged before, thereby explaining causality in ways we previously never understood. It’s important to understand that the outcomes produced as a result of these processes come from a limited universe, so this “learning” isn’t the open-ended variety which would lead to a machine becoming smarter than a human being. The machine is merely quicker and more consistent within a specific data universe.
- Artificial intelligence is the current buzzword and it creates the impression of constantly growing intelligence or intelligence that is superior to that of human beings. Many sub-sets of study and computational capability are indeed developing, from dimensionality reduction, to structured prediction to anomaly detection. But not one, or even all of these combined, is capable of leading to unbridled intelligence.
SMART(ER) ASSET MANAGEMENT
Artificial intelligence tools are making data analysis easier and more effective than ever before, and many new versions of analysing market exposure have become possible. These new quantitative methods of data processing have become mainstream over the past decade as the understanding around this technology has grown. This in-depth analysis of market exposure, however, doesn’t replace all human input, but reduces it immensely as most analysis is performed by a machine. The final portfolio construction and implementation is still managed by a portfolio manager, but one who manages according to new, modern methodologies. The portfolios designed this way are extremely cost efficient and scalable, and ultimately lead to cost savings for the investor. Tracking error or active risk is also no longer the cumulative result of what was placed into the portfolio, but rather what is placed into the portfolio is decided by the desired pre-determined tracking error. In so doing active risk can be managed accurately.
Citadel Asset Management (CAM) has been developing and using these new quantitative methods to produce custom-made sets of data for the past five years. The results have been very pleasing, so we will continue expanding the areas of application for this technology. It also forces us to remain in touch with the newest global developments in quantitative analysis as we compete against a rising number of artificial intelligence tools.
This approach also positions us well to answer the all-important question of whether machines will replace people. The answer is, simply, no. Machines, algorithms and artificial intelligence will merely replace certain research functions and analytics. However, it will create greater demand for other, higher intelligence levels of human input. In other words it will drive the need for talent and skills.
Machines won’t replace people, but they will replace mundane and repetitive tasks. These technologies will also put to shame those who base their investment hypotheses on basic or superficial analysis, since new data-rich analysis will unearth superior insight. At CAM we embrace technology and the ability it gives us to understand and create superior risk-return optimised portfolios at the lowest possible cost for our clients.
We fully embrace the new era of computer-assisted product design and portfolio management, but at Citadel we firmly believe that good, old-fashioned experience won’t soon be programmed into a machine or turned into an algorithm. This is exactly why you’ll still be able to look your Citadel portfolio manager in the eye for quite some time to come.
Written by George Herman, Citadel Chief Investment Officer