2 - Misperceptions Surrounding Artificial Intelligence in Alternative Investment Portfolio Management
Update: 2022-12-21
Description
Angelo Calvello, PhD discusses the issues surrounding AI (Artificial Intelligence) in Alternative Investment portfolio management.
Transcript:
| It is clear that their denial is based upon the single fundamental and universally held belief that investing is essentially and necessarily a human activity. Here's a common expression of that belief in a quote someone but not identify them. It's in the paper if you want to read it. My starting point is that one way or another, investing is and will remain a fundamentally human activity even when computer driven trading represents the majority of stock market activity. |
| I am prepared to take It is axiomatic that investing will remain a fundamental human activity. Such a view of investing easily accommodates the use of traditional machine learning because this machine learning merely leverages components of human judgment at scale. It's not a replacement. It's a tool for increasing the scale and speed of investing. However, this anthropocentric view of investing cannot accept A.I. that makes possible non-human investing that autonomously learns and makes all the critical investment decisions and limits the role of humans to that of developers, not portfolio managers. |
| What is so challenging to incumbent investment managers and allocators is that this new wave of AI requires neither programing by humans to replicate the decision making process of human experts, nor deep domain knowledge of the disciplines in which it operates. Instead, through its use of deep neural networks, data and compute power, this autonomous A.I. identifies in the data themselves nonlinear statistical relationships undetectable to human based and traditional machine learning methods. |
| For example, deep learning models used in cancer diagnosis and prognosis know nothing about medicine. Yet by focusing entirely on the data, they can achieve unprecedented accuracy, which is even higher than that of general statistical applications in an colleghi. According to a review published in the Journal of Cancer Letters, it is the same with deep learning and deep reinforcement learning and investing. |
| These models know nothing about the investment canon, the CFA curriculum value momentum. They're not programed to mimic the decision making of the greatest human investors. Instead, these algorithms just hunt through the data identifying patterns and similarities between the target and the data, and then use this knowledge to make investment predictions or decisions. An instructive example of such powerful self-learning algorithms is DeepMind Alphazero, which initially was developed to play the extremely complex board game Go. |
| Unlike IBM's Deep Blue, a human designed human engineered, hard coded computer program built in the 1992, played a much simpler game of chess. Alphazero started tabula rasa without human data or engineering and with no domain knowledge beyond the rules of the game. It uses a novel form of reinforcement learning in which Alphazero becomes its own teacher. The system starts off with a neural network that knows nothing about the game of Go, and it then plays games against itself. |
| Millions of games by combining this neural network with a powerful search algorithm as it plays, the neural network is tuned and updated and it predicts its moves even better. And over the course of millions of games of self playing, the system progressively learns the game of golf from scratch, accumulating thousands of years of human knowledge during a period of just a few days. |
| Alphazero also discovered new knowledge, developing unconventional strategies and creative new moves. And while many argue that deep reinforcement learning like this may be good at board games, they argue its success can be generalized to other domains. However, the unprecedented success of these and other experiments led the DeepMind team to draw the general conclusion that reinforcement learning can be used to achieve superhuman results in other domains. |
| This is a quote from a paper they published. Our results comprehensively demonstrate that a pure reinforcement learning approach is fully feasible, even in the most challenging domains, it is possible to train to a superhuman level without human experience or guidance. Given no knowledge of the domain beyond basic rules. At the end of the quote, In the face of this peer reviewed and highly cited research and other similarly robust research, investment managers and allocators continue to staunchly claim that investing is always a human process. |
| This is the exact point in the argument where good quants should provide an abundance of empirical evidence in support of their claim of the impossibility of modern human investing. Yet none is offered. Instead, the investor case and suppositions appeals to tradition and strawman arguments. |
Comments
In Channel
Download from Google Play
Download from App Store
United States00:00
00:00
x
