Artificial intelligence (AI) is a hot topic. Innovations in this area are creating intrigue as to whether machines will replace human beings as portfolio managers.
Artificial intelligence (AI) is a hot topic. Innovations in this area are creating intrigue as to whether machines will replace human beings as portfolio managers. ‘Robo-advisers’ already exist, providing basic financial advice using algorithms that are calculated from the questions being asked.
Despite recent advances, many applications of this area of computer science are still in their infancy and this is particularly the case when attempting to apply AI to asset management. It is our belief that the real source of value to the industry is Intelligence Augmentation (IA), which uses AI to help humans make decisions, rather than making the decisions for them.
One type of AI is machine learning – the use of statistical algorithms and techniques to learn the patterns in large volumes of data. One of the most common commercial applications of machine learning is predictive analytics to help forecast what future results might be. AI systems like IBM’s Watson and Deep Blue and Google’s AlphaGo have been highly publicised successes in the field of games (for Jeopardy!, Chess and Go respectively).
While AI systems winning games make for great headlines, in practice the real world is far more complicated. Many AI systems produce incorrect outputs and almost all of them require some additional human intervention – coding and algorithm amendments – in order to function properly in real world settings.
What determines the quality of learning is the quality of the input. When it comes to obtaining optimal AI outputs, there are five key conditions for success:
- A constant environment where the rules are fixed and don’t change
- The relevant information is digital, quantified
- Abundant amounts of data
- Low uncertainty
- Clear objectives
These parameters are met in games like Chess or Go due to the nature of the universe: the rules are set, there is little uncertainty around the rules and there is one clear objective. Almost limitless amounts of data can be created by making the computer play against itself. We believe that this is not the case for fundamental investing, where these five conditions are hardly ever met. It is our view that long-term investing will remain a human task, because the conditions for AI to be effective are lacking.
IA, on the other hand, has been around for about the same time as AI, has a strong track record and can be observed in virtually all areas of our daily lives. IA is simply the enhancement of intelligence through technological means.
For the asset management industry, IA is a much more relevant area of science than AI. It enables the extraction of insights that few others can even identify – even with the data being there in plain sight. This has tremendous advantages when it comes to fundamental investing.
Any fund manager considering their investments has access to many useful pieces of information about a company – its financial state, its revenues, the stated plans of its management. But there are other important things that investors do not currently have access to through traditional channels.
The datasets that contain this information are far too large and unstructured for an investment professional to utilise using conventional methods such as Excel. This is where AI and machine learning can be useful tools to assist turning data into the insight to fill a blind spot, thus augmenting the intelligence of the fund manager. However, the individual fund manager will remain in control of investment decisions.
Source: Mark Ainsworth – Head of Data Insights and Analytics, Schroders (5/2/2019)